TWM608616U - Intelligent image security identification system - Google Patents

Abstract

This creation provides an intelligent image security identification system including: a camera device and a terminal device, the camera device captures the original image in the manufacturing and processing environment; the terminal device stores a plurality of standard operation models, and each standard operation model is an operation corresponding to a different processing operation Personnel posture; the terminal device generates a person image model from the original image through image processing, and compares the person image model with the matched standard operating model to determine whether the person image model is safe for processing, and when the person image model is judged to be unsafe for processing , The terminal device generates a warning signal; through image processing combined with artificial intelligence identification, the operation behavior of the on-site operator is monitored, and a warning is generated when the operation action is inaccurate to avoid dangerous behaviors, so as to achieve the effect of intelligent monitoring safety.

Description

Intelligent image security identification system

This creation is related to a field of image recognition, especially an intelligent image security recognition system that can be used in a manufacturing and processing environment to determine whether the operator's actions are safe when operating a machine.

Among the deaths, disability, injuries and disasters in the entire industry, industrial disasters and injuries account for the highest proportion. Among them, if you analyze by industry, it can be found that the injuries and disasters in the metal product manufacturing industry are even higher; Further analysis, among the entire industry manufacturing industry, the metal manufacturing industry has the highest injury rate caused by heavy machinery.

According to the occupational accident investigation, although the factory’s machinery and equipment are usually equipped with safety precautions, such as fences, safety buttons, or fences and other fool-proof devices, in order to seek more efficient production capacity, workers sometimes deliberately use manual work to release Safety protection measures led to the final panic in the rush to work, causing an irreversible tragedy.

Due to the large number of workers on the production line, relying solely on human resources to monitor these dangerous behaviors is not only impractical in practice, but also increases the burden of management labor costs.

In order to solve the above problems, this creation provides an intelligent image security identification system, through image processing combined with artificial intelligence identification, to monitor the operating behavior of on-site operators, and the operating actions are not correct. Real-time warnings are generated to avoid dangerous behaviors, so as to achieve the effect of intelligent monitoring and safety.

An embodiment of the present creation provides an intelligent image security identification system, which includes: a camera device installed in a manufacturing and processing environment, the camera device capturing an original image in the manufacturing and processing environment; and a terminal device, which is connected with The camera is coupled to the terminal device. The terminal device has a database and a processing module. The database has a plurality of standard operation models. Each standard operation model corresponds to the posture of the operator corresponding to different processing operations. The processing module receives the information sent by the camera. For the original image, the processing module reads each standard operating model of the database to determine the standard operating model that matches the original image; the processing module processes the original image to generate a human image model, and the processing module converts the human image model Compare with the matched standard operation model to determine whether the personnel image model is safe to process. When the personnel image model is judged to be unsafe for processing, the processing module generates a warning signal; when the personnel image model is judged to be safe for processing, process The module generates a performance analysis result.

Based on the above, this creation uses image processing combined with artificial intelligence recognition to intelligently monitor the operation behavior of the on-site operator. When the operation action is uncertain, it will instantly generate a warning signal to avoid dangerous behaviors and improve the conventional human monitoring of on-site operations. The problem of personnel, and then achieve the effect of intelligent monitoring safety.

Furthermore, this creation uses image processing combined with artificial intelligence recognition to intelligently collect operator's movement data, analyze the relationship between operator's operation and machine operation, and adjust the operation schedule based on the performance analysis result, which can effectively improve the industry's efficiency.

100: Intelligent image security identification system

10: Photography installation

20: terminal device

21: Database

22: Processing module

O: Original image

E: Environmental image

P: People image

M: Standard Operating Model

D: Field configuration diagram

I: Personnel image model

K: joint point

Figure 1 is a schematic diagram of the system architecture of this author.

Figure 2 is a flow chart of image recognition processing in this creation.

Figure 3 is the field configuration diagram of this creation.

Figure 4 is a sequence diagram of the joints of the work in this creation.

Fig. 5 is a schematic diagram of an embodiment of the original image of this creation.

Figure 6 is a schematic diagram of a human image model generated by original image processing in this creation.

In order to facilitate the description of the central idea of the present invention expressed in the column of the above-mentioned summary of the invention, specific embodiments are used to express it. The various objects in the embodiment are drawn according to the proportion, size, deformation or displacement suitable for explanation, rather than drawn according to the actual component ratio.

The direction terms mentioned in this creation, such as "up", "down", "front", "rear", "left", "right", "inner", "outer", "side", etc., are only pictures The direction of the style; therefore, the direction terms used are used to explain and understand the creation, not to limit the creation, so it is stated first.

Please refer to FIG. 1 to FIG. 6, the present creation provides an intelligent image security identification system 100, which includes: A photographing device 10 installed in a manufacturing and processing environment. The photographing device 10 captures an original image O in the manufacturing and processing environment. The original image O includes an environment image E and a person image P3, and the person image P is an operator For images in the process of processing operations, such as: loading, processing, unloading, conveying, etc., the actual personnel image P depends on various processing operations, and this creation is not limited to this.

The photographing device 10 can be one or more. In this creative embodiment, the number of the photographing device 10 is plural, and each photographing device 10 is arranged in different working areas in the manufacturing and processing environment. The number of installations of the photographing device 10 is determined according to the usage requirements, and this creation is not limited by this.

A terminal device 20, which is coupled to the photographing device 10. The terminal device 20 can be a variety of personal computers, notebook computers, smart mobile devices, or tablet computers; in this creative embodiment, the terminal device 20 is a personal computer, and the terminal device 20 and the photographing device 10 are wirelessly connected.

The terminal device 20 has a database 21 and a processing module 22. The database 21 stores a plurality of standard operation models M and a field configuration diagram D. Each standard operation model M is a posture of an operator corresponding to a different processing operation. The operation model M has a sequence diagram of operation joints (which can be regarded as a skeleton diagram of the operators corresponding to different processing operations). In this creative embodiment, forging processing is taken as an example. The standard operation model M includes reclaiming, discharging processing, and stone processing. The powder spraying and processing are discharged, as shown in Figure 4, and the standard operation model M can be established in the database 21 according to different manufacturing and processing processes. This creation is not limited to this; the field configuration diagram D is for each photograph The working area corresponding to the device 10 is shown in FIG. 1 and FIG. 3.

Please refer to FIGS. 1 to 2 and 5 to 6, the processing module 22 receives the original image O transmitted by the photographing device 10, the processing module 22 receives the continuous original image O, and the processing module 22 passes through the region of interest The encoding technology processes the original image into a fixed area and a dynamic tracking area, the fixed area corresponds to the environment image E, and the dynamic tracking area corresponds to the person image P; then, the processing module 22 converts the person image P through the image matrix conversion to generate a matrixed image , The processing module 22 processes the coordinates of each joint point by matrix image processing, and the processing module 22 performs smoothing processing and image denoising processing on the coordinates of each joint point to generate a person image model I, wherein the processing module 22 Mark a joint point K on the human image model I according to the coordinates of each joint point.

In order to avoid the problem of skeleton flickering and occluded skeleton joints when the person image P is processed as the person image model I, the joint coordinate points will drift, the processing module 22 will successively receive the original The initial image O is gray-scaled. The pixel value of the original image O of the current frame and the pixel value of the original image O of the previous frame are calculated as a difference value, and the processing module 22 judges the posture of the operator based on the difference value and a threshold value. When the difference value does not exceed the set threshold, it is judged that the two frames of the person image P are the same. There is no need to capture the joint point coordinates again, and the joint point coordinate detection result of the current person image P is replaced with the previous Frame the joint point coordinates of the human image P; on the contrary, if the difference value exceeds the set threshold, the operator is deemed to have made a relatively large movement, and the joint point coordinate detection result of the current human image P will be maintained, and will not be replaced with The coordinates of the joint points of the person image P in the previous frame; thereby, the shaking of the output skeleton can be greatly reduced, and the quality of the output image can be improved.

Furthermore, the processing module 22 analyzes and judges the person image P of the original image O to obtain a standard operation model M that matches the person image P. The processing module 22 has a plurality of thresholds, and each threshold corresponds to a different processing. The processing module 22 judges the processing operation corresponding to the person image P of the original image O according to the threshold value, and reads the standard operation model M corresponding to the judgment from the database 21. The threshold value is the coordinate value corresponding to the different processing operation, for example: When the standard operation model M is the unloading action, the X coordinate of the operator's right shoulder joint point coordinates is used for threshold filtering. In distinguishing the feeding action and the unloading action, higher than the threshold will be regarded as the feeding action, and the lower The threshold will be regarded as a cutting action.

The processing module 22 compares and analyzes the coordinate of each joint point (joint point K) of the processed human image model I with the working joint sequence diagram of the standard operating model M to determine whether the human image model I is safe to process. Among them, judge The standard is whether the personnel image model I conforms to the standard operation model M.

When the human image model I is judged to be unsafe for processing, the processing module 22 generates a warning signal. The warning signal can be light, sound, or video. The warning signal can remind the operator to perform processing with a certain action, or can Let the monitoring personnel go to remind the operators to avoid dangerous behaviors.

The processing module 22 is provided with an analysis time, and the analysis time can be in minutes, hours, or days. The analysis time can be set according to the efficiency performance calculation of the manufacturing process, and this creation is not limited to this. When the human image model I is judged to be safe for processing, the processing module 22 calculates the processing efficiency of the operator according to the analysis time to generate a performance analysis result, that is, the processing module 22 will calculate the operator’s completion within the analysis time The number of operating procedures is used to estimate the operating efficiency of the operators and assist the management to control the on-site work conditions.

Based on the above, this creation can achieve the following effects:

1. The creative intelligent image security identification system 100 uses image processing combined with artificial intelligence identification to intelligently monitor the operation behavior of on-site operators. When the operation action is uncertain, it will instantly generate a warning signal to avoid dangerous behaviors and achieve The effect of intelligent monitoring of safety.

2. This creative intelligent image security identification system 100, through image processing combined with artificial intelligence identification, intelligently collects the operator's movement data, analyzes the relationship between the operator's operation and the machine operation, and adjusts the operation schedule according to the performance analysis result, and then Can effectively improve the efficiency of the industry.

The above-mentioned embodiments are only used to illustrate the creation, and are not used to limit the scope of the creation. All modifications or changes that do not violate the spirit of this creation belong to the scope of this creation's intention to protect.

100: Intelligent image security identification system

10: Photography installation

20: terminal device

21: Database

22: Processing module

Claims (10)

An intelligent image security identification system, which includes: A photographing device installed in a manufacturing and processing environment, the photographing device capturing an original image in the manufacturing and processing environment; and A terminal device coupled with the camera device, the terminal device having a database and a processing module, the database has a plurality of standard operation models, each standard operation model is a posture of an operator corresponding to a different processing operation; wherein , The processing module receives the original image sent by the photographing device, and the processing module reads each standard operating model of the database to determine and obtain the standard operating model matching the original image; the processing The module generates a person image model through image processing from the original image. The processing module compares the person image model with the matched standard operating model to determine whether the person image model is safe to process. When the person image model When it is judged as unsafe processing, the processing module generates a warning signal; when the person image model is judged as safe processing, the processing module generates a performance analysis result. The intelligent image safety identification system according to claim 1, wherein the processing module has an analysis time, and when the human image model is judged to be safe processing, the processing module calculates the processing efficiency of the operator according to the analysis time To produce the performance analysis result. The intelligent image safety identification system according to claim 1, wherein each of the standard operating models has an operating joint timing diagram; the personnel image model has a plurality of joint point coordinates; the processing module compares the operating joint timings of each standard operating model The graph is compared and analyzed with the coordinates of each joint point of the person's image model to determine whether the person's image model is safe to process. The intelligent image security identification system according to claim 3, wherein the processing module generates a matrixed image through the image matrix conversion of the original image, and the processing module uses the matrixed image to process the coordinates of each joint point . The intelligent image security identification system according to claim 3 or 4, wherein the processing module marks a joint point on the human image model according to the coordinates of each joint point. The intelligent image security identification system according to claim 5, wherein the processing module performs smoothing processing and image denoising processing on the coordinates of each joint point, and the coordinates of each joint point after processing are compared with the standard operating model Comparison and analysis of the time sequence diagram of the working joints to determine whether the person’s image model is safe for processing. The intelligent image security identification system according to claim 1, wherein the processing module receives the continuous original image, and the processing module performs gray-scale processing on the original image received successively, and converts the current original image The pixel value of the image before the pixel value of the original image calculates a difference value, and the processing module compares the difference value with a threshold value to determine the posture change of the operator. The intelligent image security identification system according to claim 1 or 7, wherein the original image includes an environment image and a person image; the processing module processes the original image into a fixed area and a fixed area through the region of interest coding technology The dynamic tracking area, the fixed area corresponds to the environment image, and the dynamic tracking area corresponds to the person image. The intelligent image security identification system according to claim 1, wherein the processing module is provided with a plurality of thresholds, and each threshold corresponds to a different processing operation, and the processing module determines the processing operation corresponding to the original image according to the threshold, The standard operating model corresponding to the judgment can be read from the database. The intelligent image security identification system according to claim 1, wherein there are a plurality of photographing devices, and each photographing device is not located in a different operation area in the manufacturing and processing environment; the database has a field layout map, and the field The configuration diagram is the working area corresponding to each photographing device.

Images