CN117809009A - Identity recognition method, equipment, device, computer equipment and storage medium - Google Patents

Abstract

The present application relates to an identity recognition method, apparatus, device, computer device, storage medium and computer program product. The method comprises the following steps: acquiring a first image acquired by an identity characteristic acquisition inlet aiming at identity recognition equipment, and detecting attachments of the first image to obtain a first detection result; when the first detection result represents that attachments exist at the identity characteristic acquisition inlet, triggering cleaning treatment aiming at the identity characteristic acquisition inlet; after triggering cleaning treatment, acquiring a second image acquired by aiming at the identity characteristic acquisition inlet, and detecting attachments of the second image to obtain a second detection result; and when the second detection result represents that attachments at the identity characteristic acquisition entrance are cleaned, responding to the identity characteristic image acquired from the identity characteristic acquisition entrance, and carrying out identity recognition based on the identity characteristic image. By adopting the method, the accuracy of identity recognition based on the identity characteristic image can be improved.

Description

Identity identification method, device, apparatus, computer equipment and storage medium

Technical Field

The present application relates to the field of computer technology, and in particular to an identity recognition method, equipment, device, computer equipment, storage medium and computer program product.

Background technique

With the development of computer technology, increasingly mature identity recognition technology has been widely used in various fields such as business cooperation, consumer payment, social media, security access control, etc. The implementation methods of identity recognition are becoming more and more diverse, such as identity recognition based on QR codes and identity recognition based on biometrics. Among them, identity recognition based on biometrics is the use of human inherent biological features, such as hand shape, fingerprint, face shape, retina, auricle and other biological features for identity recognition, which has become the development trend of identity recognition technology.

At present, when performing identity recognition based on QR codes and biometrics, it is often necessary to set up a camera to collect images, and perform identity recognition through the collected images. However, in different scenarios, the camera that collects images is prone to adhere to water stains, dust, etc. that affect the image collection effect. The collected images are prone to distortion problems such as blurring and missing, which reduces the accuracy of identity recognition.

Contents of the invention

Based on this, it is necessary to provide an identity recognition method, equipment, device, computer equipment, computer-readable storage medium and computer program product that can improve the accuracy of identity recognition in view of the above technical problems.

In the first aspect, this application provides an identification method. The methods include:

Obtain the first image collected for the identity characteristic collection entrance of the identity recognition device, perform attachment detection on the first image, and obtain the first detection result;

When the first detection result indicates that there is attachment at the identity feature collection entrance, the cleaning process for the identity feature collection entrance is triggered;

After triggering the cleaning process, obtain the second image collected for the identity feature collection entrance, perform attachment detection on the second image, and obtain the second detection result;

When the second detection result indicates that the attachments at the identity feature collection entrance have been cleared, in response to collecting the identity feature image from the identity feature collection entrance, identity recognition based on the identity feature image is performed.

In a second aspect, the present application also provides an identity recognition device, the device comprising: a processor, an image sensor, and an identity feature collection entrance;

The image sensor collects images through the identity feature collection entrance;

A processor configured to acquire the first image collected by the image sensor through the identity feature collection entrance, perform attachment detection on the first image, and obtain a first detection result; when the first detection result indicates the presence of attachments at the identity feature collection entrance, trigger Cleaning processing for the identity feature collection entrance; after triggering the cleaning process, obtain the second image collected for the identity feature collection entrance, perform attachment detection on the second image, and obtain the second detection result; when the second detection result represents the identity feature The attachments at the collection entrance have been cleaned, and in response to the identity feature image being collected from the identity feature collection entrance, identity recognition based on the identity feature image is performed.

In a third aspect, this application also provides an identity recognition device. The device includes:

The first image detection module is used to obtain the first image collected for the identity feature collection entrance of the identity recognition device, detect attachments on the first image, and obtain the first detection result;

An attachment cleaning module, configured to trigger a cleaning process for the identity feature collection entrance when the first detection result indicates that attachments exist at the identity feature collection entrance;

The second image detection module is used to obtain the second image collected for the identity feature collection entrance after triggering the cleaning process, perform attachment detection on the second image, and obtain the second detection result;

The identity recognition processing module is used to perform identity recognition based on the identity feature image in response to the identity feature image collected from the identity feature collection entrance when the second detection result indicates that the attachments at the identity feature collection entrance have been cleared.

In a fourth aspect, this application also provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

Obtain the first image collected for the identity characteristic collection entrance of the identity recognition device, perform attachment detection on the first image, and obtain the first detection result;

When the first detection result indicates that there is attachment at the identity feature collection entrance, a cleaning process for the identity feature collection entrance is triggered;

After the cleaning process is triggered, a second image collected at the identity feature collection entrance is acquired, and an attachment detection is performed on the second image to obtain a second detection result;

When the second detection result indicates that the attachments at the identity feature collection entrance have been cleared, in response to collecting the identity feature image from the identity feature collection entrance, identity recognition based on the identity feature image is performed.

In a fifth aspect, this application also provides a computer-readable storage medium. The computer-readable storage medium has a computer program stored thereon, and when the computer program is executed by the processor, the following steps are implemented:

Acquire a first image collected by an identity feature collection entrance of an identity recognition device, perform attachment detection on the first image, and obtain a first detection result;

When the first detection result indicates that there is attachment at the identity feature collection entrance, the cleaning process for the identity feature collection entrance is triggered;

After triggering the cleaning process, obtain the second image collected for the identity feature collection entrance, perform attachment detection on the second image, and obtain the second detection result;

When the second detection result indicates that the attachments at the identity feature collection entrance have been cleared, in response to collecting the identity feature image from the identity feature collection entrance, identity recognition based on the identity feature image is performed.

In a sixth aspect, this application also provides a computer program product. The computer program product includes a computer program that implements the following steps when executed by a processor:

Obtain the first image collected for the identity characteristic collection entrance of the identity recognition device, perform attachment detection on the first image, and obtain the first detection result;

When the first detection result indicates that there is attachment at the identity feature collection entrance, the cleaning process for the identity feature collection entrance is triggered;

After triggering the cleaning process, obtain the second image collected for the identity feature collection entrance, perform attachment detection on the second image, and obtain the second detection result;

When the second detection result indicates that the attachments at the identity feature collection entrance have been cleared, in response to the identity feature image being collected from the identity feature collection entrance, identity recognition based on the identity feature image is performed.

The above-mentioned identity recognition methods, equipment, devices, computer equipment, storage media and computer program products obtain the first image collected for the identity feature collection portal of the identity recognition device, perform attachment detection based on the first image, and determine the identity feature collection portal. When there are attachments at the identity feature collection entrance, the identity feature collection entrance is cleaned. When the attachments at the identity feature collection entrance have been cleaned, identity recognition is performed based on the identity feature image collected from the identity feature collection entrance. After cleaning the attachments present at the identity feature collection entrance, identity recognition is performed based on the identity feature image collected from the identity feature collection entrance, which can reduce the impact of attachments present at the identity feature collection entrance on the identity feature image imaging. , ensuring the imaging quality of the identity feature image, thereby improving the accuracy of identity recognition based on the identity feature image.

Description of drawings

Figure 1 is an application environment diagram of the identity recognition method in one embodiment;

Figure 2 is a schematic flow chart of an identity recognition method in an embodiment;

Figure 3 is a schematic flowchart of training an attachment detection model in one embodiment;

Figure 4 is a structural block diagram of an identity recognition device in one embodiment;

Figure 5 is a schematic diagram of the identity recognition method applied to the access control scenario in one embodiment;

Figure 6 is a schematic diagram of the identity recognition method applied to a palm payment scenario in one embodiment;

Figure 7 is a schematic flow chart of an identity recognition method in another embodiment;

Figure 8 is a schematic diagram of camera lens settings in one embodiment;

Figure 9 is a schematic flowchart of constructing a water stain detection model in one embodiment;

Figure 10 is a schematic diagram of a palm image distorted by water stains in one embodiment;

Figure 11 is another schematic diagram of the palm image being distorted by water stains in one embodiment;

Figure 12 is another schematic diagram of the palm image being distorted by water stains in one embodiment;

FIG13 is a schematic diagram of a process of performing water stain detection using a water stain detection model in one embodiment;

FIG14 is a block diagram of a structure of an identity recognition device in one embodiment;

Figure 15 is an internal structure diagram of a computer device in one embodiment.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present application more clear, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present application and are not used to limit the present application.

The identity recognition method provided by the embodiment of this application can be applied in the application environment as shown in Figure 1. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system may store data that server 104 needs to process. The data storage system can be integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can obtain the first image collected by the identity feature collection entrance of the identity recognition device, perform attachment detection on the first image, and obtain the first detection result. In the case where the first detection result indicates that there is an attachment at the identity feature collection entrance, the terminal 102 may trigger a cleaning process for the identity feature collection entrance. After triggering the cleaning process, the terminal 102 obtains the second image collected for the identity feature collection entrance, performs attachment detection on the second image, and obtains the second detection result. When the second detection result indicates that the attachments at the identity feature collection entrance have been cleared, the terminal 102 responds to collecting the identity feature image from the identity feature collection entrance and performs identity recognition based on the identity feature image. Specifically, the terminal 102 may collect the identity feature image. The identity feature image is sent to the server 104, and the server 104 performs identity recognition based on the identity feature image, and feeds back the obtained identity recognition result to the terminal 102.

In addition, the identity recognition method can also be directly implemented by the terminal 102 or the server 104 alone, that is, the terminal 102 can directly identify the collected identity feature image, or the server 104 can implement the identity recognition method alone.

Among them, the terminal 102 can be, but is not limited to, various desktop computers, laptops, smart phones, tablets, Internet of Things devices and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart vehicle-mounted devices, etc. . Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The terminal 102 may be configured with an image sensor device for image collection. The server 104 can be implemented as an independent server or a server cluster composed of multiple servers. The multiple servers involved can form a blockchain, and the server 104 can be a node on the blockchain.

In one embodiment, as shown in Figure 2, an identity recognition method is provided. The method is executed by a computer device. Specifically, it can be executed by a computer device such as a terminal or a server alone, or it can be executed by a terminal and a server together. In this article In the application embodiment, the application of this method to the terminal in Figure 1 is used as an example for explanation, including the following steps:

Step 202: Obtain the first image collected by the identity characteristic collection entrance of the identity recognition device, perform attachment detection on the first image, and obtain the first detection result.

Among them, identity recognition is the process of identifying the user's true identity information. Specifically, it can also verify whether the user's true identity information matches the identity information he claims. Identity recognition equipment is a device used for identity recognition, which can identify the user's identity so that corresponding processing can be performed based on the user's identity. For example, in an access control scenario, the identity recognition device can identify the user to confirm whether the user is a legitimate user and thus determine whether the user is allowed to enter. The identity feature collection portal is the portal where the identity recognition device collects identity features for identity recognition. When the identity features collected by the identity recognition device include images, the identity feature collection portal may be the collection portal of the image sensor of the identity recognition device, that is, the image sensor collects images through the identity feature collection portal. In specific implementation, the image sensor can include various types of cameras, and the identity feature collection entrance can be the corresponding lenses of various types of cameras.

The first image is collected for the identity feature collection portal of the identity recognition device. Specifically, it may be an image collected by the image sensor of the identity recognition device through the identity feature collection portal. Attachment detection refers to detecting whether there are attachments on the identity feature collection entrance, such as dust, water stains, stains, etc. If there are attachments on the identity feature collection entrance, when the identity recognition device collects images through the identity feature collection entrance, the attachments will interfere with the imaging, resulting in distortion problems in the collected images, such as blur, missing, occlusion, etc., which will affect Accuracy of identification. For example, when there are water droplets at the identity feature collection entrance, when the image sensor collects images, such as when the camera captures images, the water droplets at the identity feature collection entrance will also be captured, introducing imaging noise, resulting in images being captured during imaging. Water droplets interfere with the actual subject being photographed. The first detection result is a detection result obtained by detecting attachments on the first image. The first detection result may include data information such as whether there are attachments, types of attachments, and distribution areas of attachments. Based on the first detection result, it can be determined whether there is an object that will interfere with imaging attached to the identity feature collection entrance of the identity recognition device, so that it can be determined whether the identity feature collection entrance needs to be cleaned.

Specifically, the terminal can obtain the first image collected by the identity feature collection entrance of the identity recognition device, such as the terminal can receive the first image sent by the identity recognition device. In a specific implementation, the identity recognition device itself can directly implement the identity recognition method, that is, the identity recognition device can be used as a terminal to obtain the first image collected by the identity feature collection entrance, and perform attachment detection on the first image to obtain a first detection result. When performing attachment detection on the first image, the terminal can perform attachment detection on the first image based on target detection algorithms, such as sliding window algorithms (Sliding Window Algorithm), candidate region algorithms (Region Proposal Algorithms), selective search algorithms (Selective Search), R-CNN (Region-Convolutional Neural Networks, regional convolutional neural networks) and other algorithms to obtain a first detection result.

Step 204: When the first detection result indicates that there is an attachment at the identity feature collection entrance, a cleaning process for the identity feature collection entrance is triggered.

Among them, the cleaning process for the identity feature collection entrance refers to the process of removing the attachments present at the identity feature collection entrance. Through the cleaning process, the attachments at the identity feature collection entrance can be removed to ensure that the identity characteristics collection entrance passes through the cleaning process. The imaging quality during image acquisition means better image acquisition effect and clearer images.

Specifically, if the first detection result indicates that there is an attachment at the identity feature collection entrance, that is, continuing to collect images through the identity feature collection entrance will cause the obtained image to be distorted, then the terminal triggers the cleaning process of the identity feature collection entrance, that is, the terminal Remove attachments at the entrance to identity collection. In a specific implementation, the terminal can control the cleaning device to clean the identity feature collection entrance. For example, if the attachment at the identity feature collection entrance is water stains, the terminal can control the water stain removal device to remove the water stains, such as drying through a drying device, blowing and evaporating through a fan, etc. In addition, different types of attachments can correspond to different cleaning processing methods, and specifically different cleaning devices can be controlled to perform cleaning processing. If the attachment at the identity feature collection entrance is dust or stains, the terminal can control the scrubbing device to scrub the identity feature collection entrance, thereby removing the dust or stains at the identity feature collection entrance.

Step 206: After triggering the cleaning process, obtain the second image collected for the identity feature collection entrance, perform attachment detection on the second image, and obtain the second detection result.

The second image is an image captured by the identity recognition device at the identity feature acquisition entrance after the cleaning process is triggered, that is, the second image is an image captured again through the identity feature acquisition entrance after the attachment is cleaned.

Specifically, after the cleaning process is triggered, that is, after the attachments present at the identity feature collection entrance are removed, the terminal acquires the second image collected for the identity feature collection entrance. The terminal performs attachment detection again on the second image and obtains a second detection result. Based on the second detection result, it can be determined whether there are still objects that will interfere with imaging at the identity feature collection entrance of the identity recognition device, so that it can be determined whether the attachments have been cleared. In a specific application, for the detection of attachments in the second image, the same detection method as the detection of attachments in the first image can be adopted to ensure the consistency of the detection methods.

Step 208: When the second detection result indicates that the attachments at the identity feature collection entrance have been cleared, in response to collecting the identity feature image from the identity feature collection entrance, perform identity recognition based on the identity feature image.

Among them, the identity feature image is an image collected through the identity feature collection entrance for identity recognition, and may specifically include biometric feature images, credential feature images, etc. For example, identity feature images may include palm part images, face images, QR code images, etc.

Specifically, if the second detection result indicates that the attachments at the identity feature collection entrance have been cleaned, that is, the attachments at the identity feature collection entrance have been removed and the subsequent image collection effect can be ensured, then the terminal is based on the identity feature collection entrance. Identity recognition is performed on the identity feature images collected here. In a specific application, when the second detection result indicates that there is no attachment at the identity feature collection entrance, the terminal can obtain the identity feature image collected from the identity feature collection entrance, and perform identity recognition based on the identity feature image. There is no longer interference from attachments at the entrance to the identity feature collection in the identity feature image, and the imaging quality is high. Identity recognition based on the identity feature image can ensure the accuracy of identity recognition.

In addition, when the second detection result indicates that the attachments at the identity feature collection entrance have not been cleaned, that is, the attachments at the identity feature collection entrance have not been removed after the cleaning process, the terminal can trigger cleaning of the identity feature collection entrance again. Process until it is determined that the attachments at the identity feature collection entrance have been cleared, and then the identity feature image is collected through the identity feature collection entrance for identity recognition. In specific applications, the terminal can record the number of cleaning processes. When the number of cleaning processes reaches the threshold and the attachments have not been cleaned, the cleaning process for the identity feature collection entrance can be ended to avoid repeated cleaning for a long time. The attachments still cannot be removed even after treatment. In addition, when the attachments at the identity feature collection entrance have not been cleaned, the terminal can also generate prompt information and send the prompt information to the control end to prompt the controller at the control end to clean the identity feature collection entrance to ensure the identity features. Attachments at the collection entrance can be removed immediately. In specific implementation, a threshold value for the number of cleaning processes can be set. When the number of cleaning processes does not exceed the threshold value, if the attachments at the identity feature collection entrance have not been cleaned up after cleaning, the cleaning process can be repeated. The cleaning process ends until the attachments are cleaned or the number of times of cleaning exceeds the threshold. Statistical analysis can also be performed on the historical records of the cleaning process. For example, when cleaning the attachments, the cleaning mode corresponding to the attachments can be counted, which can include cleaning time, cleaning method, cleaning intensity, etc., so that when attachments are detected, When attaching to an object, the corresponding cleaning mode can be determined based on the parameters of the attachment, such as the type and distribution range of the attachment, so that the attachment can be effectively cleaned according to the determined cleaning mode to ensure the cleaning effect.

In the above identity recognition method, the terminal obtains the first image collected for the identity feature collection entrance of the identity recognition device, performs attachment detection based on the first image, and detects attachments based on the identity feature collection entrance when it is determined that there is an attachment at the identity feature collection entrance. The entrance is cleaned, and when the attachments at the identity feature collection entrance have been cleaned, identity recognition is performed based on the identity feature image collected from the identity feature collection entrance. After cleaning the attachments present at the identity feature collection entrance, identity recognition is performed based on the identity feature image collected from the identity feature collection entrance, which can reduce the impact of attachments present at the identity feature collection entrance on the identity feature image imaging. , ensuring the imaging quality of the identity feature image, thereby improving the accuracy of identity recognition based on the identity feature image.

In one embodiment, performing attachment detection on the first image to obtain the first detection result includes: performing image feature extraction on the first image to obtain the image features of the first image; and obtaining the identity feature collection entrance and the first image. The included distance parameters between the collection objects are included; attachment detection is performed based on the image features and distance parameters to obtain the first detection result.

The image features of the first image are obtained by extracting image features from the first image. Specifically, they can be obtained by processing the first image through various target detection algorithms. Based on the image features of the first image, it can be determined whether there is an attachment at the identity feature collection entrance of the identity recognition device. The collection object is an object included in the first image, specifically the shooting target in the first image. The identity recognition device shoots the collection object through the identity feature collection entrance to obtain the first image. The distance parameter may include the distance between the identity feature collection entrance and the shooting target. Specifically, it may include the distance from the camera lens to the shooting target. The distance parameter can be determined by a distance sensor set in the identity recognition device, such as a light distance sensor. When there is attachment at the identity feature collection entrance of the identity recognition device, specifically when there is liquid attachment, the liquid attachment itself will form a lens, which will cause abnormalities in the detection of the distance parameter. The distance parameter can assist in determining the location of the identity recognition device. Check whether there are attachments at the entrance to the identity feature collection.

Specifically, the terminal extracts image features from the first image. The terminal can extract image features from the first image through various image feature extraction methods, such as through SIFT (Scale-Invariant Features Transform), SURF Feature extraction algorithms such as Speeded Up Robust Features and HOG (Histogram of Oriented Gradient) extract image features from the first image. In addition, the terminal can also perform image feature extraction on the first image based on an artificial neural network algorithm, such as various pre-trained artificial neural network models, to obtain the image features of the first image. The terminal obtains a distance parameter between the identity feature collection entrance and the collection object included in the first image, and the distance parameter can be detected by a light distance sensor.

The terminal performs attachment detection based on the image features and distance parameters of the first image to obtain a first detection result. The terminal can perform attachment detection based on the image features and distance parameters respectively, and then combine the detection results of the two to obtain a first detection result. The terminal can also directly combine the image features and distance parameters to uniformly perform attachment detection to obtain a first detection result. In specific implementation, the terminal can perform target detection based on the image features of the first image to detect whether there are attachments at the identity feature collection entrance. For example, when the first image has distortion problems such as blur, occlusion, and missing, it may be caused by the presence of attachments at the identity feature collection entrance. The terminal can perform anomaly analysis based on the distance parameter to determine whether the distance parameter is abnormal. If the distance parameter is abnormal, it may be caused by the presence of attachments at the identity feature collection entrance. The terminal uses the image features and distance parameters of the first image to perform attachment detection, which can determine whether there are attachments at the identity feature collection entrance from multiple dimensions, thereby improving the accuracy of attachment detection.

In addition, the terminal can also perform attachment detection based on the image features of the first image alone to obtain the first detection result, that is, the terminal can only use the image features of the first image to detect attachments. Further, if attachment detection is performed solely based on the image features of the first image, and the obtained first detection result indicates the presence of attachments at the identity feature collection entrance, the terminal triggers a cleaning process for the identity feature collection entrance. If the first detection result indicates that there is no attachment at the identity feature collection entrance, the terminal can further obtain the distance parameter between the identity feature collection entrance and the collection object included in the first image, and perform attachment detection based on the distance parameter, This ensures the accuracy of attachment detection through a two-level detection method. The terminal can also use the image features and distance parameters of the first image to weight the attached objects. Specifically, the terminal can use the image features and distance parameters of the first image to detect attached objects respectively, and combine the two according to the preset weight parameters. The attachment detection results are weighted to obtain the first fused detection result. Among them, the weight parameters can be set based on the historical records of attachment detection to distinguish the importance of image features and distance parameters in attachment detection.

In this embodiment, the terminal performs attachment detection based on the image features extracted from the first image and the distance parameter between the identity feature collection entrance and the collection object included in the first image, thereby detecting attachments from features in multiple dimensions. Detecting whether there are attachments at the identity feature collection entrance can improve the accuracy of attachment detection.

In one embodiment, the first image includes a visible light image and an infrared image collected for the same collection object; performing image feature extraction on the first image to obtain the image features of the first image includes: performing separate operations on the visible light image and the infrared image. Image feature extraction, obtaining the visible light image features of the visible light image and the infrared image features of the infrared image.

Among them, the visible light image is a color image collected by a visible light image sensor, and the infrared image is an image collected by an infrared image sensor. Visible light images and infrared images are collected for the same collection object. A visible light image sensor and an infrared image sensor can be installed in the identity recognition device. When the image collection is triggered, the visible light image sensor and the infrared image sensor can collect images at the same time, so that the visible light image and the infrared image are respectively collected for the same collection object. Both visible light images and infrared images can be used for identification. Visible light image features are image features extracted from visible light images, and infrared image features are image features extracted from infrared images.

Specifically, the terminal performs image feature extraction on the visible light image and the infrared image respectively to obtain the visible light image features of the visible light image and the infrared image features of the infrared image. The visible light image features and the infrared image features can be used as image features of the first image. In specific implementation, visible light images and infrared images have different image characteristics, and different image features can be extracted through different feature extraction algorithms respectively, which is beneficial to ensuring the accuracy of image features.

Further, detecting attachments based on image features and distance parameters to obtain the first detection result includes: feature fusion of visible light image features, infrared image features and distance features of distance parameters to obtain detection features; detecting attachments based on the detection features Test and get the first test result.

Among them, the distance feature is based on the feature determined by the distance parameter. Specifically, the distance parameter can be obtained by feature mapping. For example, the terminal can perform linear or nonlinear mapping processing on the distance parameter to obtain the distance feature of the distance parameter. The detection features are obtained through feature fusion of visible light image features, infrared image features and distance features, and can characterize whether there are attachments at the identity feature collection entrance of the identity recognition device.

Specifically, the terminal obtains the distance feature of the distance parameter, and the distance feature can be obtained by the terminal performing feature mapping on the distance parameter. The terminal fuses visible light image features, infrared image features and distance features. The specific terminal can first connect the visible light image features, infrared image features and distance features, and fuse the connected features to obtain detection features. In specific applications, different weights can be set for visible light image features, infrared image features and distance features, and the visible light image features, infrared image features and distance features can be weighted and fused according to their respective weights to obtain detection features. The respective weights of visible light image features, infrared image features and distance features can be set in advance, and different weights can be set under different environmental conditions. For example, in environmental conditions with strong lighting, if the imaging quality of visible light images is high and the detection effect of visible light images is good, the weight of visible light image features can be increased; while in environmental conditions with weak lighting, the infrared image features can be improved weight to ensure the expressiveness of the detection features. The terminal detects attachments based on the detection features. For example, the terminal can detect attachments based on the detection features through a pre-selected target detection algorithm to obtain the first detection result. In a specific implementation, the terminal may input detection features into a pre-trained attachment detection model to detect attachments based on the input detection features through the attachment detection model, and output the first detection result from the attachment detection model. The first detection result may describe whether there is any attachment at the identity feature collection entrance of the identity recognition device.

In this embodiment, the first image includes a visible light image and an infrared image. The terminal performs attachment detection based on the detection features obtained by fusing the visible light image features, the infrared image features and the distance feature of the distance parameter, which can make full use of the features of different types of images, and Detecting attachments based on the distance between the collection object and the identity feature collection entrance can improve the accuracy of attachment detection.

In one embodiment, attachment detection is performed on the first image to obtain a first detection result based on an attachment detection model; attachment detection is performed on the second image to obtain a second detection result based on an attachment detection model; as shown in FIG3 , the training steps of the attachment detection model include:

Step 302: Obtain the sample image and the distance sample parameters between the identity feature collection entrance and the collection objects included in the sample image; the sample image and distance sample parameters carry attachment detection labels.

Among them, the attachment detection for the first image and the second image is realized by a pre-trained attachment detection model, that is, the first image and the second image can be respectively input into the attachment detection model, and the attachment detection model performs attachment detection respectively and outputs the corresponding detection results. The sample image is an image historically collected through the identity feature collection entrance, and the distance sample parameter is the distance parameter between the identity feature collection entrance and the collection object included in the sample image for each sample image. The sample image and the distance sample parameter both carry attachment detection tags, and the attachment detection tags are used to mark whether there are attachments at the identity feature collection entrance.

Specifically, the training of the attachment detection model can be performed by a computing device, which can be a terminal or a server. The computer device acquires a training sample, including a sample image and a distance sample parameter between the identity feature collection entrance and the collection object included in the sample image, and the sample image and distance sample parameter carry an attachment detection label.

Step 304: extract image features from the sample image using the attachment detection model to be trained to obtain sample image features of the sample image.

Among them, the sample image features are image features extracted from the sample image. Specifically, the computer device performs image feature extraction on the sample image through the attachment detection model to be trained. Specifically, the sample image can be input into the attachment detection model, and the attachment detection model performs image feature extraction to obtain a sample image of the sample image. feature.

Step 306 , using the attachment detection model to be trained, based on the sample image features and the distance sample parameters, attachment detection is performed to obtain a sample detection result.

Specifically, the computer device performs attachment detection through the attachment detection model to be trained, and the attachment detection model to be trained performs attachment detection based on the obtained sample image features and distance sample parameters to obtain the sample detection result output by the attachment detection model to be trained.

Step 308: Based on the sample detection results and attachment detection labels, the attachment detection model to be trained is updated and the training is continued until the training is completed, and the trained attachment detection model is obtained.

Specifically, the computer device updates the attachment detection model to be trained based on the sample detection results and the attachment detection labels. The computer device can compare the sample detection results with the attachment detection labels, and detect the attachments to be trained based on the comparison results. The model parameters in the model are adaptively adjusted to realize the training update of the attachment detection model to be trained. The computer equipment continues training based on the updated attachment detection model. Specifically, the training can be performed through the next sample image until the end of the training. For example, when the number of training times reaches the training times threshold, or when the accuracy of attachment detection reaches the accuracy threshold, the computer equipment End the training and obtain the completed attachment detection model. The trained attachment detection model can detect attachments on the input image and output attachment detection results on the input image.

In this embodiment, the computer device trains the attachment detection model through sample images and distance sample parameters, and performs attachment detection through the trained attachment detection model, which can ensure the accuracy of attachment detection and the detection processing efficiency.

In one embodiment, the sample image includes a visible light sample image and an infrared sample image; image feature extraction is performed on the sample image through the attachment detection model to be trained, and the sample image features of the sample image are obtained, including: through the attachment detection model to be trained The model performs image feature extraction on the visible light sample image and the infrared sample image respectively, and obtains the visible light sample image features of the visible light sample image and the infrared sample image features of the infrared sample image.

Among them, the visible light sample image is a color sample image collected by a visible light image sensor, and the infrared sample image is a sample image collected by an infrared image sensor. The visible light sample image and the infrared sample image are collected from the same collection object.

Specifically, the computer device extracts image features from the visible light sample image and the infrared sample image through the attachment detection model to be trained, and obtains visible light sample image features of the visible light sample image and infrared sample image features of the infrared sample image. The visible light sample image features and the infrared sample image features can be used as sample image features of the sample image. In specific implementation, the visible light sample image and the infrared sample image can be input into the attachment detection model to be trained, and feature extraction is performed by different feature extraction layers in the attachment detection model to be trained.

Further, through the attachment detection model to be trained, attachments are detected based on the sample image features and distance sample parameters, and the sample detection results are obtained, including: performing feature mapping on the distance sample parameters through the attachment detection model to be trained, and obtaining the distance Sample features; feature fusion of visible light sample image features, infrared sample image features and distance sample features to obtain detection sample features; perform attachment detection based on detection sample features to obtain sample detection results.

The distance sample feature is based on the feature determined by the distance sample parameter, and can be obtained by performing feature mapping on the distance sample parameter, such as performing feature mapping on the distance sample parameter through the feature mapping layer in the attachment detection model to be trained to obtain the distance sample feature. The detection sample feature is obtained by fusing the visible light sample image feature, the infrared sample image feature and the distance sample feature, and can characterize whether there is attachment at the identity feature collection entrance of the identity recognition device.

Specifically, the computer device performs feature mapping on the distance sample parameters through the attachment detection model to be trained. For example, the distance sample parameters can also be input into the attachment detection model to be trained, so as to use the attachment detection model to be trained. The feature mapping layer performs feature mapping on the distance sample parameters to obtain distance sample features. The computer device fuses the visible light sample image features, infrared sample image features and distance sample features through the attachment detection model to be trained. Specifically, the feature fusion layer in the attachment detection model to be trained can fuse the visible light sample image features , infrared sample image features and distance sample features are feature fused to obtain detection sample features. The computer device performs attachment detection based on the characteristics of the detection sample through the attachment detection model to be trained. For example, the attachment detection layer in the attachment detection model to be trained can be used to detect attachments on the characteristics of the detection sample and output the obtained sample detection. result. The sample detection results are used to describe whether there are attachments at the identity feature collection entrance when collecting sample images through the identity feature collection entrance of the identity recognition device.

In this embodiment, the sample images include visible light sample images and infrared sample images. By training the attachment detection model with the visible light sample images, the infrared sample images, and the corresponding distance sample parameters, the features of different types of images and the distance features between the collection object and the identity feature collection entrance can be fully utilized to train the attachment detection model, which can improve the accuracy of attachment detection through the attachment detection model.

In one embodiment, the identity recognition method further includes: acquiring historical identity recognition data of the identity recognition device; and determining recognition statistical parameters according to the historical identity recognition data.

Among them, the historical identity recognition data is the data of the identity recognition device's historical identity recognition, and the historical identity recognition data may record the trigger time and identity recognition results of each identity recognition performed by the identity recognition device. The recognition statistical parameters are obtained by statistically analyzing the historical identity recognition data, and may specifically include the identity recognition success rate, identity recognition duration, etc. The identity recognition success rate can be obtained based on the proportion of successful identity recognition times each time the identity recognition device performs identity recognition; the identity recognition duration can be obtained based on the time statistics of each time the identity recognition device performs identity recognition. The recognition statistical parameters can reflect the sensitivity of identity recognition based on the identity recognition device. The higher the identity recognition success rate and the shorter the identity recognition duration, the higher the sensitivity of the identity recognition device.

Specifically, the terminal obtains historical identity recognition data of the identity recognition device. Specifically, the terminal may obtain historical identity recognition data according to different dimensions. For example, according to the time dimension, the terminal can obtain historical identification data within a certain time range; or according to the number dimension, the terminal can obtain historical identification data for the most recent preset number of identifications. The terminal determines identification statistical parameters based on historical identification data. Specifically, the terminal can analyze the historical identification data, count the number of successful identifications in the historical identification data, and obtain the identification success rate; by counting the time spent on each identification , get the identification duration. The identification statistical parameters are used to characterize the sensitivity of the identification device in historical identification processing. The higher the sensitivity, the more efficiently and accurately the identification device can perform identification processing.

Further, when the first detection result indicates that there is an attachment at the identity feature collection entrance, triggering a cleaning process for the identity feature collection entrance, including: when the first detection result indicates that there is an attachment at the identity feature collection entrance, and the identification statistical parameters satisfy Cleaning trigger condition triggers the cleaning process for the identity feature collection entrance.

Among them, the cleaning trigger condition is used to determine whether there is an attachment at the identity feature collection entrance, so as to determine whether to trigger the cleaning process for the identity feature collection entrance. Cleaning trigger conditions can be adaptively set based on identification statistical parameters according to actual needs. For example, the cleaning trigger condition may be that the identification success rate is lower than the success rate threshold, or the change in the identification success rate reaches the success rate change threshold. The cleaning trigger condition can also be that the identification duration reaches the preset duration, or the change in the identification duration reaches the duration change threshold.

Specifically, the terminal compares the recognition statistical parameters with the preset cleaning trigger conditions, and determines whether the recognition statistical parameters meet the cleaning trigger conditions. If the recognition statistical parameters meet the cleaning trigger conditions, and the first detection result indicates that there is an attachment at the identity feature collection entrance. If there is something attached, it is considered that there is indeed an attachment at the identity feature collection entrance, and the terminal triggers the cleaning process for the identity feature collection entrance. For example, the first detection result indicates the presence of attachments at the identity feature collection entrance, and the identity recognition success rate based on historical identity recognition data is lower than the success rate threshold, indicating that the attachments present at the identity feature collection entrance have affected identity recognition. If the identity recognition success rate of the device affects the sensitivity of the identity recognition device, the terminal triggers the cleaning process of the identity feature collection entrance. In specific applications, the terminal can periodically detect whether there are attachments at the identity feature collection entrance, and periodically determine whether the identification statistical parameters meet the cleaning trigger conditions. When detecting the presence of attachments or identification statistical parameters at the identity feature collection entrance, When the cleaning trigger condition is met, it may be triggered to determine whether the trigger condition of another cleaning process is met. For example, when it is first detected that there is an attachment at the identity feature collection entrance, the terminal can further determine whether the identification statistical parameters meet the cleaning trigger conditions, and if so, trigger the cleaning process for the identity feature collection entrance; for another example, if the identification is first detected When the statistical parameters meet the cleaning trigger conditions, the terminal can further determine whether there are attachments at the identity feature collection entrance, and if so, trigger cleaning of the identity feature collection entrance.

In this embodiment, when the first detection result indicates the presence of attachments at the identity feature collection entrance, and the identification statistical parameters determined based on historical identity recognition data meet the cleaning trigger conditions, the terminal determines that there are attachments at the identity feature collection entrance. , it needs to be processed, the terminal triggers the cleaning process for the identity feature collection entrance, and historical identity recognition data can be combined to accurately determine attachments.

In one embodiment, when the first detection result indicates the presence of attachments at the identity feature collection entrance, triggering a cleaning process for the identity feature collection entrance includes: when the first detection result indicates the presence of attachments at the identity feature collection entrance, through the identity The attachment cleaning device of the identification equipment cleans the identity feature collection entrance.

Among them, the attachment cleaning device is a device for cleaning the identity feature collection entrance, and may specifically include a drying device, a scrubbing device, etc. When there are attachments at the identity feature collection entrance, the attachment cleaning device can be activated to clean the identity feature collection entrance to clean the attachments at the identity feature collection entrance.

Specifically, when the first detection result indicates that there are attachments at the identity feature collection entrance, the terminal can activate the attachment cleaning device of the identity recognition device, and clean the identity feature collection entrance through the attachment cleaning device. For example, the terminal can start a drying device or a scrubbing device to dry or scrub the identity feature collection entrance, thereby cleaning the identity feature collection entrance and removing the attachments at the identity feature collection entrance. In specific applications, different types of attachments can be cleaned by different types of attachment cleaning devices. Furthermore, different parameters of the attachments can be cleaned by different cleaning modes. Among them, the parameters of the attachments may include but are not limited to the type and distribution range of the attachments; the cleaning mode may include but is not limited to the cleaning device, cleaning method, cleaning time, cleaning intensity, etc.

In this embodiment, the terminal uses the attachment cleaning device of the identity recognition device to clean the identity feature collection entrance. The special attachment cleaning device can be used to effectively clean the attachments at the identity feature collection entrance to ensure the accuracy of the cleaning process. Effect.

In one embodiment, the attachments at the identity feature collection entrance include water stains; cleaning the identity feature collection entrance through the attachment cleaning device of the identity recognition equipment includes: controlling the activation of infrared lights of the identity recognition equipment to pass through The infrared light cleans water stains at the entrance to identity collection.

Among them, water stains refer to water droplets at the identity feature collection entrance. Water stains attached to the identity feature collection entrance will affect the imaging effect of images collected through the identity feature collection entrance, thereby affecting the accuracy of identity recognition. Infrared lights include lamps that emit infrared light. Infrared lights can also provide supplementary light for infrared image sensors to improve the imaging effect of infrared images. The infrared light can be set at the relevant position of the identity feature collection entrance, so that it can supplement the light when collecting images at the identity feature collection entrance, and can also clean up the water stains attached to the identity feature collection entrance.

Specifically, the attachments attached to the identity feature collection entrance include water stains. When cleaning the water stains, the terminal can control the infrared light of the identity recognition device to start, so as to clean the water stains at the identity feature collection entrance through the irradiation of the infrared light. Specifically, the infrared light can dry the water stains to accelerate the evaporation of the water stains, thereby achieving the cleaning of the water stains. In specific implementation, when cleaning the water stains at the identity feature collection entrance by starting the infrared light, the terminal can keep detecting the water stains, and when it is detected that the water stains have been cleaned, the infrared light can be turned off to end the cleaning process.

In this embodiment, the terminal controls the infrared light of the identity recognition device to start, so as to clean the water stains at the identity feature collection entrance through the infrared light, which can safely clean the water stains without contact and ensure the cleaning effect of the water stains. .

In one embodiment, the identity recognition method further includes: when the second detection result indicates that no attachment is detected, determining that the attachment at the identity feature collection entrance has been cleared.

Among them, the second detection result indicates that no attachment is detected, that is, the attachment is detected based on the second image, indicating that there is no attachment at the identity feature collection entrance, and it can be determined that the original attachment has been cleaned. Specifically, when the second detection result indicates that no attachment is detected, the terminal determines that the attachment at the identity feature collection entrance has been cleared, that is, there is no attachment at the identity feature collection entrance.

In this embodiment, when the second detection result indicates that no attachment is detected, the terminal determines that the attachment has been cleaned, which can ensure the imaging effect of subsequent images collected through the identity feature collection portal.

In one embodiment, the identity recognition method further includes: determining the attachment distribution area based on the second detection result, and determining the attachment at the identity feature collection entrance when the attachment distribution area is smaller than the attachment distribution area determined based on the first detection result. Cleaned.

Among them, the attachment distribution area refers to the area where attachments are distributed at the identity feature collection entrance. The larger the attachment distribution area, the larger the surface area of the attachments attached at the identity feature collection entrance, and generally the greater its impact on the imaging effect. Specifically, the terminal determines the attachment distribution area based on the second detection result. When the second detection result indicates that attachments still exist, the terminal can determine the attachment distribution area of the attachment at the identity feature collection entrance based on the second detection result. The terminal compares the attachment distribution area determined based on the second detection result with the attachment distribution area determined based on the first detection result, and compares the size relationship of the area ranges. If the attachment distribution area determined based on the second detection result is smaller than the attachment distribution area determined based on the first detection result, it is considered that although the attachment has not been completely cleaned, the attachment is still cleaned to a certain extent to reduce its distribution. According to the scope of the area, the terminal can determine that the attachments at the entrance to the identity collection have been cleared.

In this embodiment, when the attachment distribution area determined based on the second detection result is smaller than the attachment distribution area determined based on the first detection result, the terminal determines that the attachment has been cleared, which can improve the efficiency of subsequent image collection through the identity feature collection portal. imaging effect.

In one embodiment, in response to collecting the identity feature image from the identity feature collection entrance, performing identity recognition based on the identity feature image includes: in response to the identity recognition triggering event, acquiring the identity feature collected from the identity feature collection entrance. Image; match the identity feature image with the pre-stored registered identity information to perform identity recognition based on the identity feature image.

Among them, the identity recognition triggering event refers to an event that triggers identity recognition, which may specifically include but is not limited to operations, instructions, etc. that trigger identity recognition. For example, in the access control system scenario, when the user needs to pass the access control, the identity recognition event is triggered; another example is, when the user pays at the payment terminal, the identity recognition event is triggered. In addition, identity recognition can also be used in anti-addiction system scenarios. For example, in an online game anti-addiction system, it is necessary to limit the online gaming time of minors. When the anti-addiction is triggered, such as the accumulated time of online gaming by game users. When the preset duration threshold is reached, the game user needs to be identified. At this time, an identification event is triggered to determine whether the game user is an adult or whether he or she is the person with the game account, thereby limiting the online gaming time of minors.

In the specific implementation, the identity recognition triggering event is an event that triggers identity recognition through biometrics. Biometrics are the biometric features of the user's measurable body parts, such as hand shape, fingerprint, face shape, iris, retina, palm and other types of biometrics. feature. When performing identity recognition processing through the biometric characteristics of the user's measurable body parts, it is necessary to collect biological data from the user's body parts and extract the biometric characteristics from the collected biological data, so as to identify the user based on the extracted biometric characteristics. . For example, if the identity recognition triggering event triggers identity recognition through face, the terminal needs to collect face data from the user's face, and identify the user based on the collected face data, such as face images; another example is , if the identity recognition triggering event triggers identity recognition through the palm, the terminal needs to collect palm data from the user's palm, and identify the user based on the collected palm data. The registered identity information is the identity information entered by the user when registering his or her identity in advance, which may specifically include a registration characteristic image.

Specifically, when an identity recognition triggering event is detected, such as when a user is detected to perform identity recognition, the terminal responds to the identity recognition triggering event and obtains the identity feature image collected from the identity feature collection entrance. The terminal queries the pre-stored registered identity information and matches the identity feature image with the registered identity information. Specifically, the identity feature image can be matched with the registered feature image to perform identity recognition based on the identity feature image. Specifically, the identity feature image can be matched based on the identity feature image. The image feature matching result between the image and the registered feature image determines the identity recognition result based on the identity feature image.

In this embodiment, after the attachments at the identity feature collection entrance have been cleaned, the terminal responds to the identity recognition trigger event and matches the identity information with the registered identity information based on the identity feature image collected through the identity feature collection entrance, thereby achieving Identity recognition based on identity feature images can reduce the impact of attachments present at the identity feature collection entrance on the identity feature image imaging, ensure the imaging quality of the identity feature image, and thereby improve the accuracy of identity recognition based on identity feature images.

In one embodiment, the identity feature image is an image collected for the palm part; the registered identity information includes palm print registration features and palm vein registration features obtained by identity registration of the registered user's palm; the identity feature image is compared with the pre-stored registration Identity information is matched with identity information to perform identity recognition based on identity feature images, including: extracting palmprint features and palm vein features from identity feature images; matching palmprint features with palmprint registration features to obtain palmprint features. The palm vein feature matching result is obtained; the palm vein feature is matched with the palm vein registration feature to obtain the palm vein feature matching result; the identity recognition result is obtained based on the palm print feature matching result and the palm vein feature matching result.

Among them, the identity feature image is an image collected from the palm part, that is, the identity is recognized through the user's palm. Palmprint registration features are the palmprint features entered when a registered user registers their identity through the palm of their hand; palm vein registration features are the palm vein features entered when a registered user registers their identity through the palm of their hand.

Palmprint refers to the palm image from the end of the fingers to the wrist, which includes various features such as main lines, wrinkles, small textures, ridge ends, bifurcation points, etc. that can be used for identification. Palmprint features refer to the features reflected by the texture information of the palm, which can be extracted from the palm image by taking an image of the palm. Different users generally correspond to different palmprint features, that is, the palms of different users have different texture features. Based on the palmprint features, the identity recognition processing of different users can be realized. Palm vein refers to the vein information image of the palm, which is used to reflect the vein image information in the human palm. It has the ability to identify living objects and can be captured by an infrared camera. Palm vein features are the vein features of the palm obtained based on palm vein analysis. Different users generally correspond to different palm vein features, that is, the palms of different users have different vein features. Based on the palm vein features, the identity of different users can also be realized. identification processing. The palmprint feature matching result is the matching result obtained by feature matching based on palmprint features, which reflects the recognition result of identity recognition through palmprint. The palm vein feature matching result is the matching result obtained by feature matching based on the palm vein feature, which reflects the recognition result of identity recognition through the palm vein.

Specifically, the terminal can perform feature extraction on the identity feature image to obtain palmprint features and palm vein features. In specific applications, the identity feature image is an image collected from the palm, which can include visible light images and infrared images. The terminal extracts features from visible light images to obtain palmprint features, and the terminal extracts features from infrared images to obtain palm vein features. The terminal performs palmprint feature matching with palmprint registration features to obtain a palmprint feature matching result. In specific implementation, palmprint feature matching can calculate palmprint feature similarity, thereby obtaining a palmprint feature matching result including palmprint similarity. If the palmprint similarity exceeds the palmprint similarity threshold, the palmprint matching can be considered consistent, otherwise the palmprint matching is considered inconsistent. The terminal matches palm vein features with palm vein registration features to obtain palm vein feature matching results. In specific implementation, palm vein feature matching can calculate palm vein feature similarity, thereby obtaining a palm vein feature matching result including palm vein similarity. If the palm vein similarity exceeds the palm vein similarity threshold, the palm vein matching can be considered consistent; otherwise, the palm vein matching is considered inconsistent. The terminal obtains the identity recognition result based on the palmprint feature matching result and the palm vein feature matching result. For example, the terminal can perform weighted fusion on the palmprint feature matching result and the palm vein feature matching result, so as to obtain the identity recognition result based on the weighted fusion result.

In this embodiment, feature matching is performed on the palm print features and palm vein features to achieve identity recognition, and accurate identity recognition can be performed based on the palm image.

In one embodiment, the identity recognition method further includes: during the process of triggering the cleaning process, in a perceptible manner, providing waiting prompt information for instructing the user to wait for identity recognition through the identity recognition device.

Among them, perceptible methods refer to methods that can be intuitively perceived by users, such as playing sounds through speakers, displaying content through display screens, and providing instructions through lights. The waiting prompt information is used to instruct the user to wait. After the cleaning process for the identity feature collection entrance is completed, identity recognition through the identity recognition device can be supported. The content form of the waiting prompt information can be flexibly configured according to actual needs. For example, it can be one of text, pictures, graphics, video, and audio, or a combination of multiple forms of content. Specifically, during the process of triggering the cleaning process, the terminal may provide waiting prompt information instructing the user to wait for identity recognition through the identity recognition device in a manner perceptible to the user. For example, the cleaning waiting prompt information can be displayed on the display screen of the identification device, or the waiting prompt information can be broadcast through the speaker voice.

In this embodiment, during the process of triggering the cleaning process, the terminal provides waiting prompt information in a perceptible manner to instruct the user to wait for identity recognition through the identity recognition device. The user can be reminded in time to wait for the cleaning process to be completed to avoid the user triggering the cleaning process. The user experience can be ensured if identification failure occurs during the identification process.

In one embodiment, obtaining the first image collected by the identity feature collection portal of the identity recognition device includes: when the detection trigger condition is met, obtaining the first image collected by the identity feature collection portal of the identity recognition device; and meeting the detection trigger condition. , including at least one of reaching the attachment detection time, detecting an identity recognition trigger event, or the recognition statistical parameters of the identity recognition device meeting the statistical parameter trigger conditions.

Among them, the detection trigger condition is used to determine whether it is necessary to trigger the detection of attachments at the identity feature collection entrance, that is, whether it is necessary to detect whether there are attachments at the identity feature collection entrance. The attachment detection time is a set time for attachment detection that is preset according to actual needs. When the attachment detection time is reached, the detection triggering conditions are considered to be met. An identity recognition trigger event refers to an event that triggers identity recognition. When an identity recognition trigger event is detected, it can be considered that the detection trigger condition is met. The identification statistical parameters are obtained by collecting statistics on historical identification data, and may specifically include identification success rate, identification duration, etc. The statistical parameter trigger condition is used to determine whether attachment detection processing is required for the identity feature collection entrance based on the recognition statistical parameters. For example, satisfying the statistical parameter trigger condition can be that the identification success rate is lower than the success rate threshold, or the change in the identification success rate reaches the success rate change threshold. The statistical parameter trigger condition may also be met when the identification duration reaches a preset duration, or when the change in the identification duration reaches a duration change threshold.

Specifically, the terminal obtains a pre-set detection trigger condition and determines whether the detection trigger condition is met. When the detection trigger condition is met, it indicates that the identity feature collection entrance needs to be processed for attachment detection. The terminal obtains the first image collected for the identity feature collection entrance of the identity recognition device to perform attachment detection on the identity feature collection entrance based on the first image. In a specific application, when the attachment detection time is reached, the identity recognition trigger event is detected, or the identification statistical parameters of the identity recognition device meet the statistical parameter trigger condition, it can be considered that the detection trigger condition is met, and the attachment detection for the identity feature collection entrance is triggered. When the first detection result based on the first image indicates that there is no attachment at the identity feature collection entrance, it can return to continue to determine the detection trigger condition, and when the detection trigger condition is met next time, the attachment detection processing for the identity feature collection entrance is triggered again.

In this embodiment, when the attachment detection time is reached, the identity recognition triggering event is detected, or the identification statistical parameters of the identity recognition device meet the statistical parameter trigger conditions, the attachment detection processing for the identity feature collection entrance is triggered, and the identity features can be detected in a timely manner. Detecting attachments at the collection entrance can promptly detect attachments at the identity feature collection entrance and clean them, thereby reducing the impact of attachments at the identity feature collection entrance on the identity feature image imaging and ensuring the accuracy of the identity feature image. Imaging quality, thereby improving the accuracy of identity recognition based on identity feature images.

In one embodiment, the identity recognition method further includes: determining resource transfer parameters in response to a resource transfer triggering event; determining a target resource account based on the identity recognition result of the identity feature image; and performing resource transfer on the target resource account based on the resource transfer parameters.

Among them, resources are assets that can be exchanged for subject matter. Resources can be funds, electronic vouchers, shopping coupons, virtual red envelopes, etc. The virtual red envelope is a virtual object with certain financial numerical attributes. For example, funds can be exchanged for goods of equal value after transactions are made. Resource transfer refers to the exchange of resources, including resource transfer parties and resource transfer parties. Resources are transferred from resource transfer parties to resource transfer parties. For example, in the payment process of shopping, funds are transferred as resources. Resource transfer triggering events refer to events that trigger resource transfer, which may include but are not limited to operations, instructions, etc. that trigger resource transfer. The resource transfer trigger event can be triggered by the user who needs to perform resource transfer processing. For example, it can be triggered by the resource transfer party in the resource transfer process, or it can be triggered by the resource transfer party in the resource transfer process. Resource transfer is the transfer of resources. A certain amount of resources held by the outgoing party is transferred to the resource transferring party. The resource transfer triggering event can be flexibly set according to actual needs. The resource transfer parameters are the relevant parameters corresponding to the resource transfer triggering event to be processed, which may include but are not limited to resource transfer party, resource transfer party, resource transfer amount, discount amount, order number, resource transfer time, resource Transfer various parameter information such as terminal. The target resource account is the resource account associated with the user who triggered the resource transfer triggering event. By performing the resource transfer operation on the target resource account, resource transfer processing for the user can be implemented.

Specifically, the terminal can determine the resource transfer parameters in response to the resource transfer triggering event, such as determining the resource transfer amount, resource transfer party, etc. If the user identity corresponding to the user can be determined based on the identity recognition result, the terminal can determine the target resource account associated with the user based on the identity recognition result. Specifically, the terminal can determine the user identity corresponding to the user based on the identity recognition result, and determine the user identity corresponding to the user based on the identity recognition result. , determine the target resource account associated with the user, and the target resource account includes the user's resources. The terminal transfers resources to the target resource account based on the determined resource transfer parameters. For example, according to the resource transfer amount in the resource transfer parameters, the resources in the target resource account are transferred to the resource transfer party in the resource transfer parameters, thereby achieving Resource transfer processing for users.

In this embodiment, the target resource account is determined based on the identity recognition result, and when the resource transfer triggering event is triggered, the resource transfer process is performed based on the determined target resource account according to the corresponding resource transfer parameters, and the resource transfer process is performed based on the identity recognition result. Improved the processing efficiency of resource transfer.

In one embodiment, as shown in FIG. 4 , an identity recognition device 400 is provided, including: an image sensor 402 , an identity feature collection entry 404 , and a processor 406 ; wherein:

The image sensor 402 collects images through the identity feature collection entrance 404;

The processor 406 is used to acquire the first image collected by the image sensor through the identity feature collection entrance, perform attachment detection on the first image, and obtain the first detection result; when the first detection result indicates the presence of attachments at the identity feature collection entrance, Trigger the cleaning process for the identity feature collection entrance; after triggering the cleaning process, obtain the second image collected for the identity feature collection entrance, perform attachment detection on the second image, and obtain the second detection result; when the second detection result represents the identity The attachments at the feature collection entrance have been cleaned, and in response to the identity feature image collected from the identity feature collection entrance, identity recognition based on the identity feature image is performed.

Among them, the image sensor 402 is used for image collection to perform identity recognition through the collected images. For different identity characteristic images collected, the image sensor 402 may correspond to different types of sensors. For example, image sensor 402 may include a visible light image sensor, an infrared image sensor, or the like. The identity feature collection portal 404 is a portal for the identity recognition device 400 to collect identity features for identity recognition, that is, the image sensor 402 collects images through the identity feature collection portal 404 . The processor 406 is used to detect attachments on the images collected by the image sensor 402. When the attachments at the identity feature collection entrance have been cleared, the processor 406 can perform identity recognition based on the identity feature images collected from the identity feature collection entrance. .

In the above-mentioned identity recognition device, an image sensor in the identity recognition device performs image acquisition through an identity feature acquisition entrance, a processor in the identity recognition device obtains a first image acquired by the identity feature acquisition entrance through the identity feature acquisition entrance, and attachment detection is performed based on the first image. When it is determined that attachments exist at the identity feature acquisition entrance, a cleaning process is performed on the identity feature acquisition entrance, and when the attachments at the identity feature acquisition entrance have been cleaned, identity recognition is performed based on the identity feature image collected from the identity feature acquisition entrance. After the attachments at the identity feature acquisition entrance are cleaned, identity recognition is performed based on the identity feature image collected from the identity feature acquisition entrance, which can reduce the influence of the attachments at the identity feature acquisition entrance on the imaging of the identity feature image, ensure the imaging quality of the identity feature image, and thus improve the accuracy of identity recognition based on the identity feature image.

In one embodiment, the image sensor 402 includes a visible light image sensor and an infrared image sensor; the attachments at the identity feature collection entrance 404 include water stains; the identity recognition device 400 also includes an infrared lamp; the processor 406 is also used to control the start-up of the infrared lamp to clean the water stains at the identity feature collection entrance 404 through the infrared lamp.

Among them, the visible light image sensor is used to collect visible light images, and the infrared image sensor is used to collect infrared images. The visible light image and the infrared image can be used as identity feature images for identity recognition. Water stains refer to water droplets at the identity feature collection entrance. Water stains attached to the identity feature collection entrance will affect the imaging effect of the image collected through the identity feature collection entrance, and further affect the accuracy of identity recognition. The infrared lamp includes a lamp that emits infrared light. The infrared lamp can also be used to fill in the light for the infrared image sensor to improve the imaging effect of the infrared image. The infrared lamp can be set at the relevant position of the identity feature collection entrance, so that it can fill in the light when collecting images at the identity feature collection entrance, and can also clean the attachments attached to the identity feature collection entrance, such as cleaning the water stains attached to the identity feature collection entrance.

Specifically, the identity recognition device 400 also includes an infrared light. The attachments attached to the identity feature collection entrance 404 may include water stains. When cleaning the water stains, the processor 406 may control the infrared light of the identity recognition device 400 to start, thereby The attachments at the identity feature collection entrance 404 are cleaned through the irradiation of infrared lamps. Specifically, the infrared lamp can be used to dry the water stains and accelerate the evaporation of the water stains, thereby cleaning the water stains. In specific implementation, when cleaning the water stains at the identity feature collection entrance 404 by activating the infrared light, the processor 406 can keep detecting the water stains. After detecting that the water stains have been cleaned, the processor 406 can turn off the infrared lights to end the process. Cleaning process.

In this embodiment, the processor in the identity recognition device controls the infrared light of the identity recognition device to start, so as to clean the water stains at the identity feature collection entrance through the infrared light, so as to achieve safe cleaning of the water stains without contact and ensure the cleaning effect of the water stains.

This application also provides an application scenario that applies the above-mentioned identity recognition method. Specifically, as shown in Figure 5, the application of this identity recognition method in this application scenario is as follows:

In the access control system scenario, the user can be identified through the identity recognition device. When it is determined that the user is of a legal identity, the user can be allowed to enter through the access control. Among them, the identity recognition device collects images of the user's palm and performs identity recognition based on the palm image. An image sensor can be provided in the identity recognition device to shoot the palm of the user's outstretched hand. If dust, water stains and other objects are attached to the identity feature collection entrance in the identity recognition device, that is, the collection entrance of the image sensor, the collected palm image will be distorted, affecting the identity recognition processing based on the palm image. Based on the identity recognition method in this embodiment, the identity recognition device can obtain a first image collected for the identity feature collection entrance, perform attachment detection based on the first image, and when it is determined that there are attachments at the identity feature collection entrance, clean the identity feature collection entrance, and when the attachments at the identity feature collection entrance have been cleaned, perform identity recognition based on the identity feature image collected from the identity feature collection entrance. After cleaning the attachments at the identity feature collection entrance, identity recognition is performed based on the identity feature image collected from the identity feature collection entrance, which can reduce the impact of the attachments at the identity feature collection entrance on the imaging of the identity feature image, ensure the imaging quality of the identity feature image, and thus improve the accuracy of identity recognition based on the identity feature image.

The present application also provides an application scenario, which applies the above-mentioned identity recognition method. Specifically, the application of the identity recognition method in the application scenario is as follows:

Palm brushing refers to a method of identification through the biometric features of the palm, such as palm print features and palm vein feature recognition. Palmprint recognition can identify the identity information of different users based on pictures of the palmprint area. The palm prints, palm veins and other information on the human palm are similar to the human face. They are very important biometric characteristics and are difficult to change. Therefore, identity recognition through the human palm and further resource transfer, such as shopping payment, are the key to identity recognition research. One of the trends. When performing palmprint recognition, when shooting the palm part, it is necessary to detect the palm area of the captured image. Specifically, target detection technology can be used to locate the finger seam points and extract the palm area image from the picture. In palm recognition, it is generally necessary to take color images and infrared images. The color image is a color image of natural light collected by the color sensor. It is generally used for face or palm selection, comparison recognition, etc. in face or palm payment. Infrared images are infrared images collected by pan-infrared light imaging from infrared sensors. They are generally used for live body detection and palm vein identification in face scanning or palm payment. In specific applications, you can also select the collected images. Specifically, you can select a set of color images, depth images, and infrared images that meet the prerequisites of the living body detection and contrast recognition algorithm. For example, for face recognition, you can include color images. image, depth image and infrared image; for palm recognition, it can include color images and infrared images. Specifically, color images can be selected based on face or palm angle, size, centering, and color image clarity, infrared images can be selected based on infrared image brightness, and depth images can be optimized based on depth image integrity, so that the optimal The image is processed such as body detection and contrast recognition.

However, when using palm brushing in water parks and other scenarios, it is easy to encounter water stains dripping on the camera lens, causing palm brushing to fail. Based on this, the identity recognition method provided by this embodiment is as shown in Figure 6. The user can pay by swiping his palm on the payment device. The user can extend his palm to the identity feature collection entrance on the payment device to receive payment from the payment device. Collect the user's palm image for identity recognition and then proceed with payment processing. In order to ensure the success rate of palm swiping, the payment device can identify water stains based on the attachment detection algorithm and turn on the infrared fill light to dry the water stains. The infrared fill light can be covered near the camera lens to ensure the drying effect of the water stains. . When performing attachment detection, typical water stain images can be collected and the attachment detection model can be trained to identify water stains. When water stains are identified through the attachment detection model and it is determined that the success rate of palm swiping of the payment device has declined, the infrared fill light is turned on to dry the water stains. At the same time, the user can be reminded through the user interface (UI, UserInterface) of the payment device that the device is Treat water damage. After cleaning the water stains, you can further determine in real time whether the water stains have been cleaned. After cleaning the water stains, resume brushing, identify the user and perform payment processing.

Specifically, as shown in Figure 7, the identity recognition method provided by this embodiment may include steps 702 to 708. Among them: step 702, covering and setting an infrared light at the camera lens. The payment device collects palm images through the camera for identity recognition, so that payment processing can be performed after the identity is recognized. An infrared light can be installed over the lens of the camera. On the one hand, the infrared light can be used to supplement the light of the infrared camera, and the infrared light can also be used to dry the water stains on the lens. Step 704: Collect water stain sample images and train a water stain detection model to identify water stains. The water damage detection model is trained through the collected water damage sample images, and the water damage detection model is used for water damage recognition through the trained water damage detection model. Step 706: When the water stain detection model detects water stains and the success rate of brushing is reduced, the infrared light is turned on to dry the water stains, and a user interface is used to remind that the water stains are being processed. When water stains are detected through the water stain detection model and the success rate of palm swiping is reduced, specifically if the success rate of the payment device through palm swiping is lower than the success rate threshold, it is determined that there is water stain attached to the lens and the infrared light is turned on. Dry water stains and notify the user interface that water stains are being treated. Step 708: Check whether the water stains have been cleaned, and resume brushing after the water stains are cleaned. The payment device can detect in real time whether the water stains on the lens have been cleaned. If it is determined that the water stains have been cleaned, the palm brushing process will be resumed, allowing users to pay with their palms. In specific applications, the infrared lamp settings and the training of the water stain detection model can be implemented in advance, and steps 706 and 708 are directly performed, that is, water stain detection is directly performed through the already trained water stain detection model, and the already set infrared The lamp dries the detected water stains.

Further, as shown in Figure 8, when an infrared light is set outside the camera lens, the camera includes a color camera and an infrared camera to collect color images and infrared images respectively. Around the color camera and the infrared camera, there are 2 infrared fill lights and 2 color fill lights respectively, which are used to fill in the corresponding cameras to ensure the image collection effect. In addition, around the four fill lights, there are four light distance sensors psensor (ProximitySensor), which are used to detect the distance from the shooting target to the lens. In specific applications, the distribution of cameras, fill lights and light distance sensors at the lens can be flexibly set according to actual needs.

Further, as shown in Figure 9, the process of constructing a water stain detection model includes steps 902 to 906, wherein: step 902, determine the characteristics of the water stain image, perform analysis based on the water stain image, and determine the characteristics existing in the water stain image. , in order to determine the training samples. Specifically, the water stain image includes feature 1, that is, the image becomes blurred, and the entire palm image is particularly blurred. A single image of an IR (Infrared Radiation) image or an RGB image is blurred, or both images are blurred. The blurred areas in the images are different, or there is partial blurring. As shown in Figure 10, the left side is a palm print image collected for the palm, and the right side is a palm vein image collected for the palm. As shown in the shaded part, the palm print image and palm vein image are overall blurred and distorted due to water stains. Water stain images include feature 2, that is, there are strange lights and shadows in the image, which will cause blur and distortion in some areas. As shown in Figure 11, the left side is a palm print image collected for the palm, and the right side is a palm vein image collected for the palm. As shown in the shaded area, the palm print image and palm vein image have water stains on the four fingers of the palm. Blurry distortion. Water-damaged images include characteristic 3, that is, the brightness of the image in the water-damaged area is equivalent to or not very different from the brightness in the normal area, but the image will become partially blurred. As shown in Figure 12, the left side is a palm print image collected for the palm, and the right side is a palm vein image collected for the palm. As shown in the shaded part, most areas of the palm print image and palm vein image are blurred due to water stains. distortion.

Further, in step 904, water stain sample images and distance sample data are collected and annotated. Specifically, you can randomly sprinkle some water droplets or sprays on the surface of the camera module. On the one hand, you can collect IR images, RGB images and psensor data when there are no human hands. It is found that sometimes the psensor data will be abnormal, either all abnormal or individually abnormal, while RGB The image may be blurry. On the other hand, when human hands are present, IR images, RGB images and psensor data are collected at different distances. It is found that the IR image or RGB image may be blurred, and sometimes the psensor data will be abnormal, either all abnormal or individual abnormalities, usually the maximum value. . In addition, IR images, RGB images and psensor data under normal conditions are collected to form water stain sample images and distance sample data. The obtained water stain sample images and distance sample data are annotated, and whether there is water stain in each water stain sample image and distance sample data is marked.

Further, in step 906, a water stain detection model is obtained by training the water stain sample image and the distance sample data. A deep learning model can be constructed, and relevant parameters in the deep learning model with water stains can be obtained by training the abnormal images or psensor values with water stains. In practice, the relevant data including IR images, RGB images and psensor data can be input according to the water stain detection model to determine whether water stains exist.

Further, for the water stain detection model, its input includes IR infrared image, RGB visible light image and distance parameter, which can specifically be the value of pensor. As shown in Figure 13, the infrared image is feature extracted through the first feature extraction network, the distance parameter is feature extracted through the second feature extraction network, and the visible light image is feature extracted through the third feature extraction network. Wherein, the first feature extraction network and the third feature extraction network may be CNN networks, and the second feature extraction network may be a fully connected network. The features extracted through the first feature extraction network, the second feature extraction network and the third feature extraction network are further spliced through the sub-network fusion layer. Features obtained by more than splicing are fused through the feature fusion layer. The feature fusion layer can be a fully connected network used to fuse features. The predicted value output by the feature fusion layer is detected and processed by the water stain detection layer. Specifically, it can be classified by softmax and then two outputs are obtained, which are the probability of water beads and the probability of no water beads. Based on the probability of water droplets and the probability of no water droplets, it can be determined whether there is water stains on the lens of the payment device.

Furthermore, when the payment device recognizes water stains through the attachment detection model and determines that the success rate of palm brushing of the payment device has declined, it turns on the infrared fill light to dry the water stains, and at the same time, the user device can be reminded through the user interface of the payment device. Water damage is being treated. After cleaning the water stains, you can further determine in real time whether the water stains have been cleaned. After cleaning the water stains, resume brushing, identify the user and perform payment processing. When palm swiping recognition technology is applied to scenes such as water parks, palm swiping equipment can automatically identify and clean water stains to ensure the success rate of palm swiping recognition.

It should be understood that although the steps in the flowcharts involved in the above-mentioned embodiments are shown in sequence as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated in this article, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least some of the steps in the flowcharts involved in the above embodiments may include multiple steps or stages. These steps or stages are not necessarily executed at the same time, but may be completed at different times. The execution order of these steps or stages is not necessarily sequential, but may be performed in turn or alternately with other steps or at least part of the steps or stages in other steps.

Based on the same inventive concept, the embodiment of the present application also provides an identity recognition device for implementing the identity recognition method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more identity recognition device embodiments provided below can refer to the limitations on the identity recognition method above, and will not be repeated here.

In one embodiment, as shown in Figure 14, an identity recognition device 1400 is provided, including: a first image detection module 1402, an attachment cleaning module 1404, a second image detection module 1406 and an identity recognition processing module 1408, wherein :

The first image detection module 1402 is used to obtain the first image collected for the identity feature collection entrance of the identity recognition device, detect attachments on the first image, and obtain the first detection result;

The attachment cleaning module 1404 is used to trigger the cleaning process for the identity feature collection entrance when the first detection result indicates that there is attachment at the identity feature collection entrance;

The second image detection module 1406 is used to obtain the second image collected for the identity feature collection entrance after triggering the cleaning process, perform attachment detection on the second image, and obtain the second detection result;

The identity recognition processing module 1408 is configured to perform identity recognition based on the identity feature image in response to the identity feature image being collected from the identity feature collection entrance when the second detection result indicates that the attachments at the identity feature collection entrance have been cleared.

In one embodiment, the first image detection module 1402 includes an image feature extraction module, a distance parameter acquisition module and an attachment detection module; wherein: the image feature extraction module is used to extract image features from the first image to obtain the first image image features; a distance parameter acquisition module, used to obtain the distance parameter between the identity feature collection entrance and the collection object included in the first image; an attachment detection module, used to detect attachments based on image features and distance parameters, Get the first test result.

In one embodiment, the first image includes a visible light image and an infrared image collected for the same collection object; the image feature extraction module is also used to extract image features from the visible light image and the infrared image respectively to obtain visible light image features of the visible light image and infrared image features of the infrared image; the attachment detection module is also used to fuse the visible light image features, the infrared image features and the distance features of the distance parameters to obtain detection features; attachment detection is performed based on the detection features to obtain a first detection result.

In one embodiment, the attachment detection is performed on the first image, and the first detection result is obtained based on the attachment detection model; the attachment detection is performed on the second image, and the second detection result is obtained based on the attachment detection model; it also includes Training sample acquisition module, sample feature extraction module, sample detection module and model update module; among which: training sample acquisition module is used to obtain sample images and distance sample parameters between the identity feature collection entrance and the collection objects included in the sample images. ; The sample image and distance sample parameters carry attachment detection labels; the sample feature extraction module is used to extract image features from the sample image through the attachment detection model to be trained, and obtain the sample image features of the sample image; the sample detection module is used to Through the attachment detection model to be trained, attachments are detected based on sample image features and distance sample parameters to obtain sample detection results; the model update module is used to develop the attachment detection model to be trained based on the sample detection results and attachment detection labels. After updating the model, continue training until the end of the training, and obtain the trained attachment detection model.

In one embodiment, the sample image includes a visible light sample image and an infrared sample image; the sample feature extraction module is also used to extract image features from the visible light sample image and the infrared sample image through the attachment detection model to be trained, and obtain the visible light sample image. The visible light sample image features of the sample image and the infrared sample image features of the infrared sample image; the sample detection module is also used to perform feature mapping on the distance sample parameters through the attachment detection model to be trained to obtain the distance sample features; the visible light sample image features , infrared sample image features and distance sample features are feature fused to obtain detection sample features; attachment detection is performed based on the detection sample features to obtain sample detection results.

In one embodiment, it also includes an identification statistical parameter determination module, used to obtain historical identification data of the identification device; determine identification statistical parameters based on the historical identification data; and an attachment cleaning module 1404, also used to determine when the first detection result It indicates that there are attachments at the identity feature collection entrance and the recognition statistical parameters meet the cleaning trigger conditions, triggering the cleaning process for the identity feature collection entrance.

In one embodiment, the attachment cleaning module 1404 is also used to clean the identity feature collection entrance through the attachment cleaning device of the identity recognition device when the first detection result indicates that there is attachment at the identity feature collection entrance.

In one embodiment, the attachments at the identity feature collection entrance include water stains; the attachment cleaning module 1404 is also used to control the activation of the infrared light of the identity recognition device to use the infrared light to clean the water stains at the identity feature collection entrance. Cleaning process.

In one embodiment, an attachment cleaning determination module is further included, configured to determine that the attachment at the identity feature collection entrance has been cleared when the second detection result indicates that no attachment has been detected.

In one embodiment, an attachment cleaning determination module is further included, configured to determine the attachment distribution area based on the second detection result, and determine the identity feature collection entrance when the attachment distribution area is smaller than the attachment distribution area determined based on the first detection result. The attachments have been cleaned.

In one embodiment, the identity recognition processing module 1408 is also used to obtain the identity feature image collected from the identity feature collection entrance in response to the identity recognition triggering event; to match the identity feature image with the pre-stored registered identity information. , to perform identity recognition based on identity feature images.

In one embodiment, the identity feature image is an image acquired by capturing the palm; the registered identity information includes the palm print registration feature and the palm vein registration feature acquired by performing identity registration on the palm of the registered user; the identity recognition processing module 1408 is further used to extract the palm print feature and the palm vein feature from the identity feature image; perform palm print feature matching on the palm print registration feature to obtain a palm print feature matching result; perform palm vein feature matching on the palm vein registration feature to obtain a palm vein feature matching result; and obtain an identity recognition result based on the palm print feature matching result and the palm vein feature matching result.

In one embodiment, a waiting prompt module is further included, configured to provide, in a perceptible manner, waiting prompt information for instructing the user to wait for identity recognition through the identity recognition device during the process of triggering the cleaning process.

In one embodiment, the first image detection module 1402 is also used to obtain the first image collected for the identity feature collection entrance of the identity recognition device when the detection trigger condition is met; the detection trigger condition is met, including reaching attachment detection time, An identity recognition trigger event is detected or the recognition statistical parameters of the identity recognition device meet at least one of the statistical parameter trigger conditions.

In one embodiment, a resource transfer module is also included, configured to determine resource transfer parameters in response to a resource transfer triggering event; determine the target resource account based on the identity recognition result of the identity feature image; and perform resource transfer on the target resource account based on the resource transfer parameters. transfer.

Each module in the above-mentioned identity recognition device can be implemented in whole or in part by software, hardware and combinations thereof. Each of the above modules may be embedded in or independent of the processor of the computer device in the form of hardware, or may be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

In one embodiment, a computer device is provided. The computer device may be a terminal or a server. If the computer device is a terminal, its internal structure diagram may be as shown in Figure 15. The computer device includes a processor, memory, input/output interface, communication interface, display unit and input device. Among them, the processor, memory and input/output interface are connected through the system bus, and the communication interface, display unit and input device are connected to the system bus through the input/output interface. Wherein, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes non-volatile storage media and internal memory. The non-volatile storage medium stores operating systems and computer programs. This internal memory provides an environment for the execution of operating systems and computer programs in non-volatile storage media. The input/output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used for wired or wireless communication with external terminals. The wireless mode can be implemented through WIFI, mobile cellular network, NFC (Near Field Communication) or other technologies. The computer program, when executed by the processor, implements an identification method. The display unit of the computer device is used to form a visually visible picture and can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a display screen. The touch layer covered above can also be buttons, trackballs or touch pads provided on the computer equipment shell, or it can also be an external keyboard, touch pad or mouse, etc.

Those skilled in the art will understand that the structure shown in FIG. 15 is merely a block diagram of a partial structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

In one embodiment, a computer device is also provided, including a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, it implements the steps in the above method embodiments.

In one embodiment, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

In one embodiment, a computer program product is provided, including a computer program that implements the steps in each of the above method embodiments when executed by a processor.

It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all It is information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data need to comply with the relevant laws, regulations and standards of relevant countries and regions.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage. In the media, when executed, the computer program may include the processes of the above method embodiments. Any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetic variable memory (Magnetoresistive Random) Access Memory (MRAM), Ferroelectric Random Access Memory (FRAM), Phase Change Memory (PCM), graphene memory, etc. Volatile memory may include random access memory (Random Access Memory, RAM) or external cache memory. By way of illustration but not limitation, RAM can be in various forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM). The databases involved in the various embodiments provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include blockchain-based distributed databases, etc., but are not limited thereto. The processors involved in the various embodiments provided in this application may be general-purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to this.

The technical features of the above embodiments can be combined in any way. To simplify the description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, all possible combinations should be used. It is considered to be within the scope of this manual.

The above-described embodiments only express several implementation modes of the present application, and their descriptions are relatively specific and detailed, but should not be construed as limiting the patent scope of the present application. It should be noted that, for those of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all fall within the protection scope of the present application. Therefore, the scope of protection of this application should be determined by the appended claims.

Claims (20)

1. An identity recognition method, characterized in that the method includes: Obtain the first image collected for the identity characteristic collection entrance of the identity recognition device, perform attachment detection on the first image, and obtain the first detection result; When the first detection result indicates that there is attachment at the identity feature collection entrance, a cleaning process for the identity feature collection entrance is triggered; After the cleaning process is triggered, a second image captured at the identity feature acquisition entrance is acquired, and attachment detection is performed on the second image to obtain a second detection result; When the second detection result indicates that the attachment at the identity feature collection entrance has been cleared, in response to the identity feature image being collected from the identity feature collection entrance, identity recognition based on the identity feature image is performed. 2. The method according to claim 1, characterized in that the performing attachment detection on the first image to obtain a first detection result comprises: Performing image feature extraction on the first image to obtain image features of the first image; Acquire a distance parameter between the identity feature acquisition entrance and an acquisition object included in the first image; Attachment detection is performed based on the image features and the distance parameter to obtain a first detection result. 3. The method according to claim 2, characterized in that the first image includes a visible light image and an infrared image collected for the same collection object; the image feature extraction is performed on the first image to obtain the The image features of the first image include: Perform image feature extraction on the visible light image and the infrared image respectively to obtain the visible light image features of the visible light image and the infrared image features of the infrared image; The attachment detection based on the image features and the distance parameter to obtain the first detection result includes: Feature fusion is performed on the visible light image feature, the infrared image feature and the distance feature of the distance parameter to obtain detection features; Perform attachment detection based on the detection characteristics to obtain a first detection result. 4. The method according to claim 1, characterized in that: performing attachment detection on the first image and obtaining the first detection result is based on an attachment detection model; and performing attachment detection on the second image. For attachment detection, obtaining the second detection result is implemented based on the attachment detection model; the training steps of the attachment detection model include: Obtain a sample image and a distance sample parameter between the identity feature collection entrance and the collection object included in the sample image; the sample image and the distance sample parameter carry an attachment detection label; Extracting image features from the sample image using the attachment detection model to be trained to obtain sample image features of the sample image; Using the attachment detection model to be trained, attachment detection is performed based on the sample image features and the distance sample parameter to obtain sample detection results; Based on the sample detection results and the attachment detection label, the attachment detection model to be trained is updated and the training is continued until the training is completed, and the attachment detection model that has been trained is obtained. 5. The method according to claim 4, characterized in that the sample image includes a visible light sample image and an infrared sample image; the image feature extraction is performed on the sample image through the attachment detection model to be trained to obtain the The sample image features of the sample image include: Using the attachment detection model to be trained, image features are extracted on the visible light sample image and the infrared sample image respectively, and the visible light sample image features of the visible light sample image and the infrared sample image of the infrared sample image are obtained. feature; The attachment detection model to be trained is used to detect attachments based on the sample image features and the distance sample parameters to obtain sample detection results, including: Perform feature mapping on the distance sample parameters through the attachment detection model to be trained to obtain distance sample features; Perform feature fusion on the visible light sample image features, the infrared sample image features and the distance sample features to obtain detection sample features; Attachments are detected based on the detected sample characteristics to obtain sample detection results. 6. The method according to claim 1, characterized in that, the method further comprises: Obtain historical identification data of the identification device; Determine identification statistical parameters based on the historical identification data; When the first detection result indicates that there is an attachment at the identity feature collection entrance, triggering a cleaning process for the identity feature collection entrance, including: When the first detection result indicates that there is an attachment at the identity feature collection entrance and the identification statistical parameters meet the cleaning trigger condition, the cleaning process for the identity feature collection entrance is triggered. 7. The method according to claim 1, characterized in that when the first detection result indicates that there is an attachment at the identity feature collection entrance, triggering a cleaning process for the identity feature collection entrance, including: When the first detection result indicates that there is attachment at the identity feature collection entrance, the identity feature collection entrance is cleaned by an attachment cleaning device of the identity recognition device. 8. The method according to claim 7, characterized in that the attachments at the identity feature collection entrance include water stains; the attachment cleaning device passing through the identity recognition equipment is used to clean the identity feature collection entrance. Perform cleaning procedures including: The infrared lamp of the identity recognition device is controlled to start, so as to clean the water stain at the identity feature collection entrance through the infrared lamp. 9. The method according to claim 1, characterized in that the method further includes at least one of the following: When the second detection result indicates that no attachment is detected, it is determined that the attachment at the identity feature collection entrance has been cleared; The attachment distribution area is determined based on the second detection result. When the attachment distribution area is smaller than the attachment distribution area determined based on the first detection result, it is determined that the attachment at the identity feature collection entrance has been cleared. . 10. The method according to claim 1, characterized in that in response to collecting the identity feature image from the identity feature collection entrance, performing identity recognition based on the identity feature image comprises: In response to the identity recognition triggering event, obtain the identity feature image collected from the identity feature collection entrance; The identity feature image is matched with pre-stored registered identity information to perform identity recognition based on the identity feature image. 11. The method according to claim 10, characterized in that the identity feature image is an image acquired by collecting the palm part; the registered identity information includes the palm print registration feature and the palm vein registration feature acquired by performing identity registration on the palm of the registered user; Matching the identity feature image with pre-stored registered identity information to perform identity recognition based on the identity feature image includes: Extracting palm print features and palm vein features from the identity feature image; Match the palmprint features with the palmprint registration features to obtain a palmprint feature matching result; Match the palm vein features with the palm vein registration features to obtain palm vein feature matching results; According to the palmprint feature matching result and the palm vein feature matching result, an identity recognition result is obtained. 12. The method of claim 1, further comprising: During the process of triggering the cleaning process, waiting prompt information for instructing the user to wait for identity recognition through the identity recognition device is provided in a perceptible manner. 13. The method according to any one of claims 1 to 12, characterized in that said obtaining the first image collected for the identity characteristic collection entrance of the identity recognition device includes: When the detection triggering conditions are met, obtain the first image collected by the identity feature collection entrance of the identity recognition device; The satisfaction of the detection triggering condition includes at least one of reaching the attachment detection time, detecting an identity recognition triggering event, or the recognition statistical parameters of the identity recognition device meeting the statistical parameter triggering condition. 14. The method according to any one of claims 1 to 12, characterized in that the method further includes: In response to the resource transfer triggering event, determine resource transfer parameters; Determine the target resource account based on the identity recognition result of the identity feature image; Resource transfer is performed on the target resource account based on the resource transfer parameter. 15. An identity recognition device, characterized in that the device includes: an image sensor, an identity feature collection entrance and a processor; wherein, The image sensor collects images through the identity feature collection entrance; The processor is configured to acquire the first image collected by the image sensor through the identity feature collection entrance, detect attachments on the first image, and obtain a first detection result; when the first detection result represents The presence of attachments at the identity feature collection entrance triggers a cleaning process for the identity feature collection entrance; after triggering the cleaning process, a second image collected for the identity feature collection entrance is obtained, and the second image is The image is subjected to attachment detection to obtain a second detection result; when the second detection result indicates that the attachment at the identity feature collection entrance has been cleared, in response to the identity feature image being collected from the identity feature collection entrance , perform identity recognition based on the identity feature image. 16. The device according to claim 1, wherein the image sensor includes a visible light image sensor and an infrared image sensor; attachments at the identity feature collection entrance include water stains; the device further includes an infrared lamp; The processor is also used to control the activation of the infrared lamp, so as to use the infrared lamp to clean the water stains at the identity feature collection entrance. 17. An identity recognition device, characterized in that the device includes: The first image detection module is used to obtain the first image collected for the identity feature collection entrance of the identity recognition device, perform attachment detection on the first image, and obtain the first detection result; An attachment cleaning module, configured to trigger a cleaning process for the identity feature collection entrance when the first detection result indicates the presence of attachments at the identity feature collection entrance; A second image detection module, configured to obtain a second image collected for the identity characteristic collection entrance after triggering the cleaning process, perform attachment detection on the second image, and obtain a second detection result; An identity recognition processing module, configured to perform processing based on the identity feature image in response to the identity feature image being collected from the identity feature collection portal when the second detection result indicates that the attachments at the identity feature collection entrance have been cleared. Identity recognition of identity feature images. 18. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein the processor implements the steps of the method according to any one of claims 1 to 14 when executing the computer program. 19. A computer-readable storage medium with a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 14 are implemented. 20. A computer program product, comprising a computer program, characterized in that, when executed by a processor, the computer program implements the steps of the method according to any one of claims 1 to 14.