[go: up one dir, main page]

CN117058803A - An intelligent data collection method and collection system based on deep learning - Google Patents

An intelligent data collection method and collection system based on deep learning Download PDF

Info

Publication number
CN117058803A
CN117058803A CN202311324315.7A CN202311324315A CN117058803A CN 117058803 A CN117058803 A CN 117058803A CN 202311324315 A CN202311324315 A CN 202311324315A CN 117058803 A CN117058803 A CN 117058803A
Authority
CN
China
Prior art keywords
module
face
access control
digital signal
calculation
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202311324315.7A
Other languages
Chinese (zh)
Other versions
CN117058803B (en
Inventor
杨小林
刘超
付金涛
李晨
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Inspur Smart Technology Innovation Shandong Co Ltd
Original Assignee
Inspur Smart Technology Innovation Shandong Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Inspur Smart Technology Innovation Shandong Co Ltd filed Critical Inspur Smart Technology Innovation Shandong Co Ltd
Priority to CN202311324315.7A priority Critical patent/CN117058803B/en
Publication of CN117058803A publication Critical patent/CN117058803A/en
Application granted granted Critical
Publication of CN117058803B publication Critical patent/CN117058803B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/168—Feature extraction; Face representation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/172—Classification, e.g. identification
    • G—PHYSICS
    • G07—CHECKING-DEVICES
    • G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
    • G07C9/00—Individual registration on entry or exit
    • G07C9/30—Individual registration on entry or exit not involving the use of a pass
    • G07C9/32—Individual registration on entry or exit not involving the use of a pass in combination with an identity check
    • G07C9/37—Individual registration on entry or exit not involving the use of a pass in combination with an identity check using biometric data, e.g. fingerprints, iris scans or voice recognition

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Multimedia (AREA)
  • Human Computer Interaction (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Databases & Information Systems (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Collating Specific Patterns (AREA)
  • Image Analysis (AREA)

Abstract

本发明涉及一种基于深度学习的智能化数据采集方法及采集系统，涉及数据处理的技术领域，其中，数据采集方法包括初步采集、特征分析、形成特征参数、人脸信息获取、人脸特征分析、初次特征比对、人脸日常采集、初次判断、二次判断、计算等步骤；数据采集系统包括输入终端、识别终端以及门禁控制模块。本发明能够使得人员在经过门禁时的面部信息参数都与上次的进行比较，进而提高门禁对面部信息的识别精度，进而降低门禁的拒识率。

The invention relates to an intelligent data collection method and collection system based on deep learning, and relates to the technical field of data processing. The data collection method includes preliminary collection, feature analysis, formation of feature parameters, face information acquisition, and face feature analysis. , initial feature comparison, daily face collection, initial judgment, secondary judgment, calculation and other steps; the data collection system includes an input terminal, a recognition terminal and an access control module. The present invention can compare the facial information parameters of people when passing through the access control with the last time, thereby improving the recognition accuracy of the facial information by the access control, and thereby reducing the rejection rate of the access control.

Description

Intelligent data acquisition method and system based on deep learning
Technical Field
The application relates to the technical field of data processing, in particular to an intelligent data acquisition method and system based on deep learning.
Background
As an essential article for daily life of people, the entrance guard provides excellent protection experience for daily life of people. With the demands of people on intellectualization, convenience and practicability of the access control, various novel intelligent access control systems are emerging in the current market, and the access control system based on face recognition is the most popular among people. The face recognition access control system takes the face of a resident as a switch basis, can recognize resident personnel and non-resident personnel, does not depend on keys in the recognition process, and saves the practical cost of users.
At present, a region feature analysis algorithm widely adopted in the face recognition technology utilizes a computer image processing technology to extract human image feature points from video, and utilizes the principle of biometrical science to analyze and build a mathematical model, namely a face feature template. And carrying out feature analysis by using the built face feature template and the face image of the person to be tested, and giving a similarity value according to the analysis result. From this value it is determined whether the person is the same.
However, when the face information of a person is identified by the existing access control identification system, when the face information of the person is changed (such as fat and thin change, old change and the like), the access control system still compares the face information parameters according to the recorded information, so that the refusal rate of the person is improved, and the person with the recorded face information cannot pass through the access control.
Disclosure of Invention
In order to reduce the rejection rate of the access control system to personnel and improve the probability of the personnel who have recorded face information passing through the access control, the application provides an intelligent data acquisition method and an intelligent data acquisition system based on deep learning.
In a first aspect, the application provides an intelligent data acquisition method based on deep learning, which adopts the following technical scheme:
an intelligent data acquisition method based on deep learning comprises the following steps:
preliminary collection: acquiring face information and identity information of a person needing to pass through the entrance guard, acquiring a face image, and performing image size adjustment and gray scale adjustment operation on the face image;
and (3) feature analysis: performing face detection on the shot face image, positioning the position of the face, performing matching of the face object, and simultaneously performing operation by using data of a database so as to obtain a feature vector matrix of the face;
forming characteristic parameters: the feature vector matrix of the face image is converted into a first digital signal parameter M through the analysis of the face image in the feature analysis step;
face information acquisition: the access control system acquires face information of personnel passing through the access control, acquires face images, and performs image size adjustment and gray scale adjustment operation on the face images;
face feature analysis: the access control system performs face detection on the shot face image, positions the face, performs matching of the face object, and simultaneously performs operation by using data of a database, so that a feature vector matrix of the face is obtained, and the feature vector matrix of the face image is converted into a second digital signal parameter N;
primary feature comparison: setting the maximum allowable error A, ifExecuting an entrance guard opening step and a face daily acquisition step; if->The method comprises the steps of carrying out a first treatment on the surface of the Executing an access control step, and executing a preliminary acquisition step again;
daily face collection: recording face images of persons passing through the entrance guard i times, converting a feature vector matrix of the face images into a third digital signal, and recording the third digital signal as N i ；
And (3) primary judgment: if it isExecuting an entrance guard opening step, executing a secondary judging step, and executing an entrance guard closing step if not;
and (3) secondary judgment: is provided withFix the warning error B, and B satisfiesIf->Executing the calculating step;
and (3) calculating: calculating a fourth digital signal parameter Q, wherein a calculation model of the fourth digital signal parameter Q is as follows:
the method comprises the steps of carrying out a first treatment on the surface of the And m=q;
and (3) opening: controlling the opening of an access control;
door confinement: and controlling the closing of the door control.
Optionally, in the secondary judging step, the warning error B is updated at any time, and a calculation model of the warning error B is as follows:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein T is the daily face acquisition period, C is the sensitivity constant, and +.>。
Optionally, determining the value of C according to the identity information of the person in the preliminary acquisition step, where the calculation model of the value of C is:the method comprises the steps of carrying out a first treatment on the surface of the Wherein, the age is the age of the person to be collected in the preliminary collection step.
Optionally, a period threshold T1 is set, and if face information is not acquired in the period threshold T1 in the face information acquisition step, a preliminary acquisition step is performed.
In a second aspect, the present application provides a data acquisition system, which adopts the following technical scheme:
the utility model provides a data acquisition system, includes input terminal, discernment terminal and entrance guard control module, input terminal includes:
acquisition module I: the output end is in electrical signal connection with the input end of the analysis module I and is used for acquiring face information of a person needing to pass through the access control, acquiring a face image and performing image size adjustment and gray scale adjustment operation on the face image;
analysis module I: the output end is electrically connected with the input end of the computing module I and is used for carrying out face detection work on the shot face image, positioning the position of the face and carrying out matching work on the face object, and meanwhile, carrying out operation by using the data of the database so as to obtain a feature vector matrix of the face;
calculation module I: the output end is in electrical signal connection with the input end of the identification terminal, and the characteristic vector matrix of the face image is converted into a first digital signal parameter M through analysis of the face image in the analysis module I, and the first digital signal parameter M is transmitted to the identification terminal;
the identification terminal comprises an acquisition module II, an analysis module II and a comparison module I:
acquisition module II: the output end is electrically connected with the input end of the analysis module II and is used for acquiring face information of personnel passing through the entrance guard, acquiring a face image and performing image size adjustment and gray scale adjustment operation on the face image;
analysis module II: the output end is electrically connected with the input end of the computing module II, the shot face image is subjected to face detection work, the position of the face is positioned, the matching work of the face object is carried out, and meanwhile, the data of the database is used for carrying out operation, so that a feature vector matrix of the face is obtained;
calculation module II: the output end is electrically connected with the input end of the comparison module I, and the characteristic vector matrix of the face image is converted into a first digital signal parameter N through the analysis of the face image in the analysis module II;
comparison module I: the input end is electrically connected with the output end of the computing module I and the output end of the computing module II, the output end is electrically connected with the input end of the access control module, and the maximum allowable error A is obtained and used for comparing the second digital signal parameter N with the first digital signal parameter M;
the access control module: and controlling the opening and closing of the access control system according to the comparison result in the comparison module I.
Optionally, the identification terminal further comprises an acquisition module III, a judgment module I, a judgment module II and a calculation module;
acquisition module III: the input end is electrically connected with the output end of the comparison module I, the output end is electrically connected with the input end of the judgment module I, and the comparison module I is used for recording face images of I times of personnel passing through the entrance guard, converting a feature vector matrix of the face images into a third digital signal and recording the third digital signal as N i；
Judging module I: the output end is electrically connected with the input end of the access control module and the input end of the judging module II and is used for outputting a third digital signal N i Make a judgment whenWhen the door control system is in a closed state, the judging module transmits a judging result to the door control module and the judging module II;
judging module II: the output end is electrically connected with the input end of the calculation module, the warning error B is input in the judgment module II, and B meets the requirementIf->When the judgment result is received, the judgment result is transmitted to a calculation module;
the calculation module: a fourth digital signal parameter Q is calculated,and allows M to be overlaid.
Optionally, the judgment module II also updates the warning error B in real time through a formulaDetermining the value of B, wherein T is the daily face acquisition period, C is the sensitivity constant, and +.>。
Optionally, determining the value of C according to the identity information of the person in the acquisition module I, where the value calculation model of C is:the method comprises the steps of carrying out a first treatment on the surface of the Wherein, the age is the age of the person to be collected in the preliminary collection step.
Optionally, the period threshold T1 is included in the digital data in the acquisition module II, and if the acquisition module II does not acquire the face information in the period of the period threshold T1, the signal is transmitted to the acquisition module I.
In summary, the present application includes at least one of the following beneficial technical effects:
1. and when the person passes through the entrance guard, the maximum allowable error is set, so that whether the face data of the person passing through the entrance guard exceeds the range of the maximum allowable error is judged, and the probability that the person cannot pass through the entrance guard system when the face of the person is slightly decorated is reduced, and the rejection rate of the bedding entrance guard system is further reduced.
2. When the face information is input into the access control system, the access control system can recognize the face information of the person entering and exiting the access control according to the input face, and the face of the person can be changed along with the increase of time, so that the access control system records the face information N of the person passing through the access control for i times by setting the warning error value B i Judgment of N i Whether the face information parameters are in the range of the warning error B or not, and reassigning the initial face information parameters M, so that the face information parameters of a person when passing through the entrance guard are compared with the face information parameters of the last time, the recognition accuracy of the entrance guard on the face information is improved, and the refusal rate of the entrance guard is reduced.
3. Because the face information of young people and old people changes greatly, the face information of young people changes less, and the value of C is determined according to people in preliminary information acquisition by setting the sensitivity constant C, so that the accuracy of the face recognition of the people with large changes by the access control system is improved.
Drawings
FIG. 1 is a flow chart of embodiment 1 of the present application;
FIG. 2 is a schematic diagram of a system structure according to embodiment 2 of the present application;
Detailed Description
The application is described in further detail below in connection with fig. 1-2.
Example 1: the embodiment discloses an intelligent data acquisition method based on deep learning, referring to fig. 1, the intelligent data acquisition method based on deep learning comprises the following steps:
preliminary collection: acquiring face information and identity information of a person needing to pass through the entrance guard, acquiring a face image, and performing operations such as image size adjustment, gray scale adjustment and the like;
and (3) feature analysis: performing face detection on the shot face image, positioning the position of the face, performing matching of the face object, and simultaneously performing operation by using data of a database so as to obtain a feature vector matrix of the face;
forming characteristic parameters: the feature vector matrix of the face image is converted into a first digital signal parameter M through the analysis of the face image in the feature analysis step;
face information acquisition: the access control system acquires face information of personnel passing through the access control, acquires face images, and performs operations such as image size adjustment, gray scale adjustment and the like;
face feature analysis: the access control system performs face detection on the shot face image, positions the face, performs matching of the face object, and simultaneously performs operation by using data of a database, so that a feature vector matrix of the face is obtained, and the feature vector matrix of the face image is converted into a second digital signal parameter N;
primary feature comparison: setting the maximum allowable error A, ifExecuting an entrance guard opening step and a face daily acquisition step; if->The method comprises the steps of carrying out a first treatment on the surface of the Executing an access control step, and executing a preliminary acquisition step again;
daily face collection: recording face images of persons passing through the entrance guard i times, converting a feature vector matrix of the face images into a third digital signal, and recording the third digital signal as N i ；
And (3) primary judgment: if it isExecuting an entrance guard opening step, executing a secondary judging step, and executing an entrance guard closing step if not;
and (3) secondary judgment: setting a warning error B, wherein B meets the following conditionIf->Executing a calculation step, wherein the warning error B is updated in real time, and a calculation model of the warning error B is as follows:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein T is the daily face acquisition period, C is the sensitivity constant, and +.>The value of C in the formula is assigned according to the crowd to which the acquired personnel belong in the preliminary acquisition step, and the value calculation model of C is as follows:the method comprises the steps of carrying out a first treatment on the surface of the Wherein, the age is the age of the person to be collected in the preliminary collection step; (e.g., if age is 20, then by the formula +.>Calculating the value of C, wherein C is about 0.866, and if the age is 50, the value of C is directly 0.998).
The calculation steps are as follows: calculating a fourth digital signal parameter Q, wherein a calculation model of the fourth digital signal parameter Q is as follows:
the method comprises the steps of carrying out a first treatment on the surface of the And M=Q (for example, the value of the initial value M is 100, the ith identification N of the entrance guard i 94, then M and N i Taking the average value as Q to be 97, enabling Q to cover M, and when the next time the access control system recognizes the face, obtaining the compared data as Q).
And (3) opening: and controlling the door control to be opened.
Door confinement: and controlling the closing of the door control.
The implementation principle of the intelligent data acquisition method based on deep learning in the embodiment is as follows: when the access control data acquisition is executed, personnel needing to pass through the access control firstly need to go to a face acquisition position to acquire faces, the personnel acquire face information and identity information of the personnel needing to pass through the access control, face images are acquired, and the operations such as image size adjustment, gray scale adjustment and the like are carried out on the face images; performing face detection on the shot face image, positioning the position of the face, performing matching of the face object, and simultaneously performing operation by using data of a database so as to obtain a feature vector matrix of the face; the feature vector matrix of the face image is converted into a digital signal parameter M through the analysis of the feature analysis on the face image, after the face information acquisition is finished, the face information is transmitted to an access control system, the access control system carries out face detection on the shot face image, positions the face, carries out matching on the face object, carries out operation by using the data of a database, and therefore the feature vector matrix of the face is obtained, the feature vector matrix of the face image is converted into a second digital signal parameter N, and the maximum allowable error A is set ifThe entrance guard is opened, and the face daily acquisition step is executed; if it isThe method comprises the steps of carrying out a first treatment on the surface of the Executing the entrance guard closing step and executing the preliminary acquisition step (re-entering the face information) again, when the entrance guard is opened, recording the face images of the personnel passing through the entrance guard i times by the entrance guard system, converting the feature vector matrix of the face images into a third digital signal, and converting the third numberThe word signal is recorded as N i If->The entrance guard is opened, a secondary judging step is executed, and if not, the entrance guard is closed; setting a warning error B, and B satisfies +.>If (if)Executing a calculation step, wherein the warning error B is updated in real time, and a calculation model of the warning error B is as follows:The method comprises the steps of carrying out a first treatment on the surface of the Wherein T (unit: month) is the daily face acquisition period, C is the sensitivity constant, andthe value of C in the formula is assigned according to the age of the person to be collected in the preliminary collection step; (e.g., if age is 20, then by the formula +.>Calculating the value of C, wherein the value of C is about 0.866, if the age is 50, the value of C is directly 0.998), and replacing the first digital signal parameter M with the fourth digital signal parameter Q, wherein the entrance guard system records the face information N of i times of personnel passing through the entrance guard i Judgment of N i Whether the face information parameters are in the range of the warning error B or not, and reassigning the initial face information parameters M, so that the face information parameters of a person when passing through the entrance guard are compared with the face information parameters of the last time, the recognition accuracy of the entrance guard on the face information is improved, and the refusal rate of the entrance guard is reduced.
Example 2: the embodiment discloses a data acquisition system, referring to fig. 2, the data acquisition system includes an input terminal, an identification terminal and an access control module.
The input terminal includes:
acquisition module I: the output end is in electrical signal connection with the input end of the analysis module I and is used for acquiring face information of a person needing to pass through the access control, acquiring a face image, and performing operations such as image size adjustment, gray scale adjustment and the like;
analysis module I: the output end is electrically connected with the input end of the computing module I and is used for carrying out face detection work on the shot face image, positioning the position of the face and carrying out matching work on the face object, and meanwhile, carrying out operation by using the data of the database so as to obtain a feature vector matrix of the face;
calculation module I: the output end is in electrical signal connection with the input end of the identification terminal, and the characteristic vector matrix of the face image is converted into a first digital signal parameter M through analysis of the face image in the analysis module I, and the first digital signal parameter M is transmitted to the identification terminal;
the identification terminal comprises an acquisition module II, an analysis module II, a comparison module I, an acquisition module III, a judgment module I, a judgment module II and a calculation module:
acquisition module II: the output end is electrically connected with the input end of the analysis module II and is used for acquiring face information of personnel passing through the entrance guard, acquiring a face image, and performing operations such as image size adjustment, gray scale adjustment and the like; the digital of the acquisition module II is provided with a period threshold T1, and if the acquisition module II does not acquire the face information in the period of the period threshold T1, a signal is transmitted to the acquisition module I.
Analysis module II: the output end is electrically connected with the input end of the computing module II, the shot face image is subjected to face detection work, the position of the face is positioned, the matching work of the face object is carried out, and meanwhile, the data of the database is used for carrying out operation, so that a feature vector matrix of the face is obtained;
calculation module II: the output end is electrically connected with the input end of the comparison module I, and the characteristic vector matrix of the face image is converted into a first digital signal parameter N through the analysis of the face image in the analysis module II;
comparison module I: the input end is electrically connected with the output end of the computing module I and the output end of the computing module II, the output end is electrically connected with the input end of the access control module, and the maximum allowable error A is obtained and used for comparing the second digital signal parameter N with the first digital signal parameter M;
acquisition module III: the input end is electrically connected with the output end of the comparison module I, the output end is electrically connected with the input end of the judgment module I, and the comparison module I is used for recording face images of I times of personnel passing through the entrance guard, converting a feature vector matrix of the face images into a third digital signal and recording the third digital signal as N i 。
Judging module I: the output end is electrically connected with the input end of the access control module and the input end of the judging module II and is used for outputting a third digital signal N i Make a judgment whenWhen the door control system is in a closed state, the judging module transmits a judging result to the door control module and the judging module II;
judging module II: the output end is electrically connected with the input end of the calculation module, the warning error B is input in the judgment module II, and B meets the requirementIf->When the judgment result is received, the judgment result is transmitted to a calculation module; the warning error B is updated in real time by the formula +.>Determining the value of B, wherein T is the daily face acquisition period, C is the sensitivity constant, and +.>Determining the value of C according to the identity information of the personnel in the acquisition module I, wherein the value calculation model of C is as follows:The method comprises the steps of carrying out a first treatment on the surface of the Wherein, the age is the age of the person to be collected in the preliminary collection step.
The calculation module: a fourth digital signal parameter Q is calculated,and allows M to be overlaid.
The access control module: and controlling the opening and closing of the access control system according to the comparison result in the comparison module I and the judgment module I.
The implementation principle of the data acquisition system of the embodiment is as follows: the acquisition module I acquires face information of personnel needing access control, acquires a face image, performs operations such as image size adjustment and gray level adjustment, and the like, then transmits the acquired and processed face information to the analysis module I, the analysis module I performs face detection on the shot face image, positions the face, performs matching of face objects, and simultaneously performs operation by using data of a database, so that a feature vector matrix of the face is obtained, the analysis module I transmits the obtained feature vector matrix information of the face to the calculation module I, and the calculation module I converts the feature vector matrix of the face image into a first digital signal parameter M through analysis of the face image in the analysis module I and transmits the first digital signal parameter M to the comparison module I in the identification terminal. When an acquired person passes through an entrance guard test for the first time, the acquisition module II acquires facial information of the person passing through the entrance guard, acquires a face image, performs operations such as image size adjustment and gray scale adjustment, and the like, transmits the information to the analysis module II, a period threshold T1 is arranged in the acquisition module II, if the acquisition module II does not acquire the face information within the period of the period threshold T1, a signal is transmitted to the acquisition module I, the analysis module II performs face detection on the shot face image, positions the face, performs matching of the face object, and performs operation by using data of a database, so that a feature vector matrix of the face image is obtained, the feature vector matrix of the face image is transmitted to the calculation module II, the calculation module II converts the feature vector matrix of the face image into a second digital signal parameter N, the comparison module I obtains a maximum allowable error A, the second digital signal parameter N shot by the entrance guard system is compared with the first digital signal parameter M, and the comparison result is transmitted to the entrance guard control module, and whether the entrance guard control module opens the door according to the entrance guard control result.
When a person passes through the entrance guard for many times, the acquisition module III records face images of the person passing through the entrance guard for i times, converts a feature vector matrix of the face images into a third digital signal, and records the third digital signal as N i The judging module I receives the third digital signal data Ni transmitted by the acquisition module and transmits a third digital signal N i Make a judgment whenWhen the warning error B is input into the judging module II, when B meets +.>If->When the warning error B is detected, the judgment result is transmitted to a calculation module, the warning error B is updated in real time, and the warning error B is judged to be the same as the warning error B according to the formula ∈>Determining the value of B, wherein T is the daily face acquisition period, C is the sensitivity constant, and +.>Determining the value of C according to the identity information of the personnel in the acquisition module I, wherein the value calculation model of C is as follows:The method comprises the steps of carrying out a first treatment on the surface of the Wherein, age is the age of the person to be collected in the preliminary collection step, and the calculation module calculates the fourth digital signal parameter Q,/-for the person to be collected>And allows M to be overlaid.
The above embodiments are not intended to limit the scope of the present application, so: all equivalent changes in structure, shape and principle of the application should be covered in the scope of protection of the application.

Claims (8)

1.一种基于深度学习的智能化数据采集方法，其特征在于：包括以下步骤：1. A deep learning-based intelligent data acquisition method, characterized by the following steps: 初步采集：对需要经过门禁的人员进行面部信息采集以及身份信息采集，获取人脸图像，并将其做图像大小调整和灰度调整操作；Preliminary data collection: Facial information and identity information are collected from personnel who need to pass through the access control system. Facial images are obtained and their size and grayscale are adjusted. 特征分析：对拍摄的人脸图像进行人脸检测工作，定位人脸的位置，进行人脸对象的匹配工作，同时运用数据库的数据进行运算，从而获得人脸的特征向量矩阵；Feature analysis: Face detection is performed on the captured face images to locate the face position, and face object matching is performed. At the same time, data from the database is used for calculation to obtain the feature vector matrix of the face. 形成特征参数：通过特征分析步骤中对人脸图像的分析，将人脸图像的特征向量矩阵转化为第一数字信号参数M；Forming feature parameters: Through the analysis of the face image in the feature analysis step, the feature vector matrix of the face image is transformed into the first digital signal parameter M; 人脸信息获取：门禁系统获取经过门禁的人员进行面部信息，并获取人脸图像，并将其做图像大小调整和灰度调整操作；Facial information acquisition: The access control system acquires the facial information of people passing through the access control system, obtains facial images, and performs image size and grayscale adjustments. 人脸特征分析：门禁系统将拍摄的人脸图像进行人脸检测工作，定位人脸的位置，进行人脸对象的匹配工作，同时运用数据库的数据进行运算，从而获得人脸的特征向量矩阵，并将人脸图像的特征向量矩阵转化为第二数字信号参数N；Facial Feature Analysis: The access control system performs face detection on captured facial images, locates the face position, performs face object matching, and uses data from the database to perform calculations to obtain the feature vector matrix of the face, and then converts the feature vector matrix of the facial image into the second digital signal parameter N; 初次特征比对：设置最大允许误差A，若，则执行门禁开步骤，并执行人脸日常采集步骤；若；执行门禁闭步骤，并再次执行初步采集步骤；Initial feature comparison: Set the maximum permissible error A, if If so, the access control opening procedure will be executed, followed by the routine facial recognition process; if Perform the access control procedure and then perform the initial data collection procedure again. 人脸日常采集：记录i次通过门禁的人员的人脸图像，并将人脸图像的特征向量矩阵转化为第三数字信号，并将第三数字信号记录为Ni；Daily face data collection: Record the face images of people passing through the access control system i times, convert the feature vector matrix of the face images into a third digital signal, and record the third digital signal as N<sub>i</sub>; 初次判断：若，则执行门禁开步骤，并执行二次判断步骤，否则执行门禁关步骤；Initial assessment: If If the condition is met, the access control opening step will be executed, followed by a secondary judgment step; otherwise, the access control closing step will be executed. 二次判断：设定警戒误差B，且B满足，若，则执行计算步骤；Secondary judgment: Set a warning error B, and B satisfies... ,like If so, then proceed with the calculation steps; 计算：对第四数字信号参数Q进行计算，第四数字信号参数Q的计算模型如下：Calculation: The fourth digital signal parameter Q is calculated. The calculation model for the fourth digital signal parameter Q is as follows: ；并另M＝Q； And let M = Q; 门禁开：控制门禁开启；Access Control Open: Controls the opening of access control doors; 门禁闭：控制门禁闭合。Access control: Controls the closing of access control doors. 2.根据权利要求1所述的一种基于深度学习的智能化数据采集方法，其特征在于：在二次判断步骤中，时刻对警戒误差B进行更新，警戒误差B的计算模型如下：2. The intelligent data acquisition method based on deep learning according to claim 1, characterized in that: in the secondary judgment step, the warning error B is updated at all times, and the calculation model of the warning error B is as follows: ；式中T为人脸日常采集的周期，式中C为灵敏度常数，且。 In the formula, T represents the daily face capture cycle, and C is the sensitivity constant. . 3.根据权利要求2所述的一种基于深度学习的智能化数据采集方法，其特征在于：根据初步采集步骤中人员的身份信息对C的取值进行确定，C的取值计算模型为：；式中age为初步采集步骤中被采集人员的年龄。3. The intelligent data acquisition method based on deep learning according to claim 2, characterized in that: the value of C is determined according to the identity information of the personnel in the preliminary acquisition step, and the calculation model for the value of C is: In the formula, age represents the age of the person being collected in the initial data collection step. 4.根据权利要求1所述的一种基于深度学习的智能化数据采集方法，其特征在于：设置周期阈值T1,若人脸信息获取步骤中在周期阈值T1时间段内未获取人脸信息，则执行初步采集步骤。4. The intelligent data acquisition method based on deep learning according to claim 1, characterized in that: a periodic threshold T1 is set, and if no face information is acquired within the periodic threshold T1 time period in the face information acquisition step, the preliminary acquisition step is executed. 5.一种数据采集系统，应用权利要求1-4中任意一项所述的基于深度学习的智能化数据采集方法，其特征在于：包括输入终端、识别终端以及门禁控制模块，所述输入终端包括：5. A data acquisition system, employing the intelligent data acquisition method based on deep learning as described in any one of claims 1-4, characterized in that: it comprises an input terminal, a recognition terminal, and an access control module, wherein the input terminal comprises: 采集模块I：输出端与分析模块I的输入端电信号连接，用于对需要经过门禁的人员进行面部信息采集，获取人脸图像，并将其做图像大小调整和灰度调整操作；Acquisition Module I: The output terminal is electrically connected to the input terminal of Analysis Module I. It is used to collect facial information of people who need to pass through the access control system, obtain facial images, and perform image size and grayscale adjustment operations on them. 分析模块I：输出端与计算模块I的输入端电信号连接，用于对拍摄的人脸图像进行人脸检测工作，定位人脸的位置，进行人脸对象的匹配工作，同时运用数据库的数据进行运算，从而获得人脸的特征向量矩阵；Analysis Module I: The output terminal is electrically connected to the input terminal of Calculation Module I. It is used to perform face detection on the captured face image, locate the position of the face, perform face object matching, and use the data in the database to perform calculations to obtain the feature vector matrix of the face. 计算模块I：输出端与识别终端的输入端电信号连接，通过分析模块I中对人脸图像的分析，将人脸图像的特征向量矩阵转化为第一数字信号参数M，并将第一数字信号参数M传输至识别终端；Calculation module I: The output end is electrically connected to the input end of the recognition terminal. Through the analysis of the face image in the analysis module I, the feature vector matrix of the face image is converted into the first digital signal parameter M, and the first digital signal parameter M is transmitted to the recognition terminal. 所述识别终端包括采集模块II、分析模块II、计算模块II、比对模块I、采集模块III、判断模块I、判断模块II以及计算模块：The identification terminal includes a data acquisition module II, an analysis module II, a calculation module II, a comparison module I, a data acquisition module III, a judgment module I, a judgment module II, and a calculation module: 采集模块II：输出端与所述分析模块II的输入端电信号连接，用于获取经过门禁的人员进行面部信息，并获取人脸图像，并将其做图像大小调整和灰度调整操作；Acquisition Module II: The output terminal is electrically connected to the input terminal of the analysis module II, and is used to acquire facial information of people passing through the access control system, acquire facial images, and perform image size and grayscale adjustment operations on them; 分析模块II：输出端与计算模块II的输入端电信号连接，将拍摄的人脸图像进行人脸检测工作，定位人脸的位置，进行人脸对象的匹配工作，同时运用数据库的数据进行运算，从而获得人脸的特征向量矩阵；Analysis Module II: The output end is electrically connected to the input end of Calculation Module II. It performs face detection on the captured face image, locates the face position, performs face object matching, and uses data from the database to perform calculations to obtain the feature vector matrix of the face. 计算模块II：输出端与比对模块I的输入端电信号连接，通过分析模块II中对人脸图像的分析，将人脸图像的特征向量矩阵转化为第一数字信号参数N；Calculation module II: The output terminal is electrically connected to the input terminal of comparison module I. Through the analysis of the face image in analysis module II, the feature vector matrix of the face image is converted into the first digital signal parameter N. 比对模块I：输入端与计算模块I的输出端以及计算模块II的输出端电信号连接，输出端与门禁控制模块的输入端电信号连接，并获得最大允许误差A，用于将第二数字信号参数N与第一数字信号参数M进行比较；Comparison Module I: Its input terminal is electrically connected to the output terminal of Calculation Module I and the output terminal of Calculation Module II, and its output terminal is electrically connected to the input terminal of the access control module. It obtains the maximum permissible error A and is used to compare the second digital signal parameter N with the first digital signal parameter M. 采集模块III：输入端与比对模块I的输出端电信号连接，输出端与判断模块I的输入端电信号连接，用于记录i次通过门禁的人员的人脸图像，并将人脸图像的特征向量矩阵转化为第三数字信号，并将第三数字信号记录为Ni； Acquisition Module III: Its input end is electrically connected to the output end of Comparison Module I, and its output end is electrically connected to the input end of Judgment Module I. It is used to record the face images of people passing through the access control system i times, convert the feature vector matrix of the face image into a third digital signal, and record the third digital signal as N i. 判断模块I：输出端与门禁控制模块的输入端以及判断模块II的输入端电信号连接，用于对第三数字信号Ni进行判断，当时，则判断模块将判断结果传递至门禁控制模块以及判断模块II；Judgment Module I: Its output terminal is electrically connected to the input terminal of the access control module and the input terminal of Judgment Module II, and is used to judge the third digital signal Ni . When this happens, the judgment module will transmit the judgment result to the access control module and the judgment module II; 判断模块II：输出端与计算模块的输入端电信号连接，在判断模块II中输入警戒误差B，且B满足，若时，则将判断结果传递至计算模块；Judgment Module II: The output terminal is electrically connected to the input terminal of the calculation module. A warning error B is input into Judgment Module II, and B satisfies... ,like If the condition is met, the judgment result will be passed to the calculation module; 计算模块：对第四数字信号参数Q进行计算， ，并使得将M进行覆盖；Calculation module: Calculates the fourth digital signal parameter Q. And thus cover M; 门禁控制模块：根据比对模块I中的比对结果，控制门禁系统的开合。Access control module: Controls the opening and closing of the access control system based on the comparison results in comparison module I. 6.根据权利要求5所述的一种数据采集系统，其特征在于：所述判断模块II中还对警戒误差B实时更新，通过公式，确定B的值，其中T为人脸日常采集的周期，式中C为灵敏度常数，且。6. The data acquisition system according to claim 5, characterized in that: the judgment module II further updates the warning error B in real time, using a formula... Determine the value of B, where T is the period of daily face capture, and C is the sensitivity constant. . 7.根据权利要求6所述的一种数据采集系统，其特征在于：根据采集模块I中人员的身份信息对C的取值进行确定，C的取值计算模型为：7. A data acquisition system according to claim 6, characterized in that: the value of C is determined based on the identity information of the personnel in acquisition module I, and the calculation model for the value of C is: ；式中age为初步采集步骤中被采集人员的年龄。 In the formula, age represents the age of the person being collected in the initial data collection step. 8.根据权利要求7所述的一种数据采集系统，其特征在于：所述采集模块II中数字有周期阈值T1，若采集模块II在周期阈值T1时段内未获取到人脸信息，则将信号传递至采集模块I。8. A data acquisition system according to claim 7, characterized in that: the digital data in the acquisition module II has a periodic threshold T1, and if the acquisition module II does not acquire face information within the periodic threshold T1 time period, the signal is transmitted to the acquisition module I.
CN202311324315.7A 2023-10-13 2023-10-13 Intelligent data acquisition method and system based on deep learning Active CN117058803B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311324315.7A CN117058803B (en) 2023-10-13 2023-10-13 Intelligent data acquisition method and system based on deep learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311324315.7A CN117058803B (en) 2023-10-13 2023-10-13 Intelligent data acquisition method and system based on deep learning

Publications (2)

Publication Number Publication Date
CN117058803A true CN117058803A (en) 2023-11-14
CN117058803B CN117058803B (en) 2024-01-05

Family

ID=88657663

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311324315.7A Active CN117058803B (en) 2023-10-13 2023-10-13 Intelligent data acquisition method and system based on deep learning

Country Status (1)

Country Link
CN (1) CN117058803B (en)

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1971630A (en) * 2006-12-01 2007-05-30 浙江工业大学 Access control device and check on work attendance tool based on human face identification technique
CN106803301A (en) * 2017-03-28 2017-06-06 广东工业大学 A kind of recognition of face guard method and system based on deep learning
CN107818308A (en) * 2017-10-31 2018-03-20 平安科技（深圳）有限公司 A kind of recognition of face intelligence comparison method, electronic installation and computer-readable recording medium
CN109670481A (en) * 2018-12-29 2019-04-23 河北三川科技有限公司 Dynamic threshold setting method is compared based on the testimony of a witness of age and gender
WO2019133996A1 (en) * 2017-12-29 2019-07-04 Brian Bertan System and method for combining identity information to facilitate image acquisition
CN115798005A (en) * 2022-11-28 2023-03-14 中国工商银行股份有限公司 Reference photo processing method and device, processor and electronic equipment
CN115830672A (en) * 2022-11-16 2023-03-21 武汉爱迪科技股份有限公司 Method, device and storage medium for constructing and updating face database based on mobile terminal
CN115830668A (en) * 2022-10-21 2023-03-21 北京声智科技有限公司 User authentication method, device, computing device and medium based on facial recognition

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1971630A (en) * 2006-12-01 2007-05-30 浙江工业大学 Access control device and check on work attendance tool based on human face identification technique
CN106803301A (en) * 2017-03-28 2017-06-06 广东工业大学 A kind of recognition of face guard method and system based on deep learning
CN107818308A (en) * 2017-10-31 2018-03-20 平安科技（深圳）有限公司 A kind of recognition of face intelligence comparison method, electronic installation and computer-readable recording medium
WO2019133996A1 (en) * 2017-12-29 2019-07-04 Brian Bertan System and method for combining identity information to facilitate image acquisition
CN109670481A (en) * 2018-12-29 2019-04-23 河北三川科技有限公司 Dynamic threshold setting method is compared based on the testimony of a witness of age and gender
CN115830668A (en) * 2022-10-21 2023-03-21 北京声智科技有限公司 User authentication method, device, computing device and medium based on facial recognition
CN115830672A (en) * 2022-11-16 2023-03-21 武汉爱迪科技股份有限公司 Method, device and storage medium for constructing and updating face database based on mobile terminal
CN115798005A (en) * 2022-11-28 2023-03-14 中国工商银行股份有限公司 Reference photo processing method and device, processor and electronic equipment

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
王越;瞿少成;陈青松;: "基于人脸识别技术的社区智能门禁系统的实现", 电子测量技术, no. 16 *

Also Published As

Publication number Publication date
CN117058803B (en) 2024-01-05

Similar Documents

Publication Publication Date Title
Zhou et al. Probabilistic human recognition from video
JP3469031B2 (en) Face image registration apparatus and method
JP4543423B2 (en) Method and apparatus for automatic object recognition and collation
CN107133612A (en) Based on image procossing and the intelligent ward of speech recognition technology and its operation method
CN109299690B (en) A method that can improve the accuracy of video real-time face recognition
TWI767775B (en) Image processing based emotion recognition system and method
CN113591535B (en) Facial feature point-based method for identifying chewing motion of old people in feeding process
CN112801000B (en) A fall detection method and system for the elderly at home based on multi-feature fusion
CN112163564A (en) Fall prediction method based on human key point behavior recognition and LSTM
Khan et al. Low dimensional representation of dorsal hand vein features using principle component analysis (PCA)
CN112395901A (en) Improved face detection, positioning and recognition method in complex environment
CN204791050U (en) Authentication equipment
CN118648868A (en) Intelligent sleep breathing state monitoring system and method
CN112307453A (en) Personnel management method and system based on face recognition
Tan et al. Face recognition algorithm based on open CV
Sadeghi et al. Modelling and segmentation of lip area in face images
CN111062308A (en) Face recognition method based on sparse expression and neural network
CN111582195B (en) A method for constructing a Chinese lip language monosyllable recognition classifier
CN110555929B (en) Laboratory entrance guard verification system and verification method
CN114360058B (en) Cross-view gait recognition method based on walking view prediction
CN117058803B (en) Intelligent data acquisition method and system based on deep learning
CN110443577A (en) A kind of campus attendance checking system based on recognition of face
CN118629075A (en) A static face recognition method and system based on feature vector method
CN119007278A (en) Intelligent behavior recognition and anomaly detection method for nursing home based on deep learning
CN111191598A (en) Facial expression recognition method based on intelligent accompanying robot

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant