WO2019104949A1

WO2019104949A1 - Residential entrance access control system which achieves human big data acquisition and analysis

Info

Publication number: WO2019104949A1
Application number: PCT/CN2018/087218
Authority: WO
Inventors: 董承利
Original assignee: 特斯联（北京）科技有限公司
Priority date: 2017-11-28
Filing date: 2018-05-17
Publication date: 2019-06-06
Also published as: CN108062809B; CN108062809A

Abstract

A residential entrance access control system that achieves human big data acquisition and analysis, comprising: an entrance access control device (101), a video surveillance device (102), an Internet of Things instrument interface device (103), a data collection device (104), and a big data aggregation and analysis platform (105). The system uses individuals entering and leaving a residential community unit as targets, extracts and identifies the faces, attire and body features of the individuals, and acquires feature information, associated information and behavioral pattern information of the individuals. The described information is aggregated in a big data platform, thus analyzing big data of information of individuals, creating portraits of the behavior of individuals, recognizing residential room features, and supporting social integration management and public services for government agencies, community properties, and like entities.

Description

A residential access control system for realizing human data collection and analysis

Technical field

The invention relates to the field of big data technology, in particular to a residential access control system for realizing large data collection and analysis of personnel.

Background technique

Personnel management is an important aspect of integrated social management. For example, government agencies need to understand the general situation of permanent residents and floating population in their jurisdictions, and provide quantitative basis for formulating policies, allocating resources, preventing public security, and maintaining social order. Units and other units are also necessary to provide information on the owners and renters of the community. The general situation of special people (such as people with disabilities, etc.) has a general understanding to strengthen the pertinence and predictability of property management and services, to match the property management with the conditions of the residential units, and to find out in time that they do not meet the living regulations. Behavior that may harm public safety.

In the traditional management methods, neighborhood committees, police stations, property companies and other institutions use manual survey methods to collect personnel information by means of door-to-door visit registration. Nowadays, the population of urban communities is increasing, the mobility is increasing, and the personnel situation is changing rapidly. The amount of information collected by the above traditional means is very limited, and the timeliness is also very lagging. Moreover, the collection efficiency is low and the labor cost is high, which is unable to adapt to the current management needs.

In the current era of big data, all walks of life require extensive data collection in as large a space as possible and as long as possible. In particular, it collects data related to every move in the daily life of the user and analyzes useful information. For example, the content of web pages browsed by users online, the items purchased by online shopping, etc., are collected as big data and used by websites to analyze user habits and preferences. Then, for the needs of social management and public services, it is necessary to apply the ideas and means of big data to achieve the normal data collection and analysis of the information of residential community personnel.

For example, Chinese patent document CN106910151 discloses a social security big data platform. The platform has a population basic database, a social security data warehouse, a data exchange platform, a social security comprehensive analysis system, a social security information management system, and a social security service portal. The system mainly uses the data exchange platform to connect the population and social security data resources of various government departments, and uses these data resources to conduct comprehensive analysis of social security and provide services such as pension and fertility in the system portal. In this prior art, the source of the big data of the personnel information is a database of government department registration management, and the system itself does not have the ability to collect and aggregate big data.

For example, Chinese Patent Document CN106886970 discloses a mobile population management system with an early warning function, including a plurality of access control hosts, a property management machine, a user terminal, a police terminal, and a server. The access control host includes an access card reader, a network module, a camera, and a security early warning module. The access control host connects to the server through the network module, and captures the face, combines the access control video with the face recognition, and finds that the suspicious individual immediately pushes the information to the public security. Department to improve community policing. The property management machine is created and managed by the user on the community and the auxiliary building, unit and house number, carries on the identity of the household and family information, binds the access card and opens the door, and newly registered resident identity and face. The information is sent to the police terminal for inquiry. In the prior art, the collection of personnel information is realized by registering personnel on the property management machine in the process of handling the access control card, and the data cannot be collected effectively for the unregistered personnel. In addition, the system utilizes access control host and other equipment, but realizes the functions of monitoring alarms and statistical unit access, and cannot provide more types of valuable information in the collection of personnel information big data, and thus cannot support diversified big data analysis. , can not achieve more rich application features.

It can be seen that in order to meet the increasingly sophisticated personnel management needs of government agencies and property units, it is necessary to implement a big data collection system for personnel information in residential communities, which can comprehensively acquire and identify various types of personnel information, especially to tap potential personnel. Manage valuable information and achieve diversified big data analytics. In order to reduce the impact on residents' normal life rhythm, it is hoped that the information collection, identification and analysis of personnel information can be completed potentially by means of more and more IoT devices deployed in residential communities.

Summary of the invention

(1) Technical problems solved

In view of the above needs of the prior art, the present invention provides a residential access control system that implements human data collection and analysis. The invention takes the personnel in the residential building as the object, and realizes the big data collection of various state information such as the personnel characteristic information, the related information and the behavior pattern information, and collects and analyzes the relevant information of the personnel for the government agencies, the community property and the like. Social integrated management and public service support.

(2) Technical plan

A residential access control system for realizing human data collection and analysis, comprising: an access control device, a video monitoring device, an Internet of Things instrument interface device, a data centralized device, and a big data convergence analysis platform;

The access control device is installed in each unit door of the residential building, and is used for door lock control, visual intercom, and monitoring the shooting of the person entering and leaving the unit door; extracting the human body image of the person entering and leaving the unit door from the captured picture a feature, the human body image feature includes facial feature information, a dress color feature, a body shape feature, and records the entry and exit time of the person according to the shooting time; and uploads the human body image feature to the data through the Internet of Things through the Internet of Things communication Concentrated device

The video monitoring device is configured to capture a person monitoring screen of a person entering the residential unit in the public space of the house, extracting a person object from the person monitoring screen, obtaining a human body image feature, a moving track information, and an associated room number of the person target, The Internet of Things is uploaded to the data concentrator;

The IoT instrument interface device is connected to the smart meter of the residential unit, and is used for periodically collecting the readings of the smart meter according to the fixed time interval, and uploading the collected meter readings to the data centralized device through the Internet of Things;

The data concentrating device is configured to cover the Internet of Things within the scope of the residential unit, and receive and store the human body image feature uploaded by the access control device, the human body image feature uploaded by the video monitoring device, the movement track information, and the associated room number through the Internet of Things. And reading the meter reading uploaded by the Internet of Things instrument interface device; identifying the human body image feature uploaded by the access control device, the human body image feature uploaded by the video monitoring device, the moving track information, the associated room number, and the Internet of Things instrument interface Correlation between meter readings uploaded by the device, integrating and storing information association registration; accessing the wide area network and transmitting the information association registration remotely to the big data convergence analysis platform;

The big data convergence analysis platform is configured to receive and store the information related registration uploaded by all data concentrating devices in the jurisdiction area through the wide area network, and apply the information related registration to perform big data analysis related to personnel management.

Preferably, the access control device comprises: an electronic door lock, a digital button, a card reader or a fingerprint recognizer, a walkie-talkie, a front camera, a rear camera, an Internet of Things communication module, and a main control chip; the electronic door lock can be in the main Locking or opening the unit door under the control of the control chip; the numeric keypad is used to input the room number and password; the main control chip starts the walkie-talkie according to the input room number, and captures the operator screen through the front camera to realize the visibility with the designated room. Voice intercom; the main control chip also compares the entered password with the preset password to implement password unlocking; the card reader or fingerprint reader can read the incoming card number held by the operator or scan the operator fingerprint, the main control The chip realizes the card unlocking or fingerprint unlocking by comparing the read in/out card number or the scanned operator fingerprint with the preset card number or the owner fingerprint information; the main control chip uses the operator image taken by the front camera to identify the current access control operation. The facial feature information of the person is compared with the preset facial feature information of the owner to realize the face The identification is unlocked; the front camera is used to capture the surveillance video of a certain space outside the unit door, from which the access control operator and the personnel entering the unit door are obtained; the rear camera is used to capture the surveillance video of a certain space within the unit door, from which the video is obtained. The screen of the person leaving the unit door; the main control chip also extracts the human body image features of the person entering and leaving the unit door from the pictures taken by the front camera and the rear camera, and records the entry and exit time of the person according to the shooting time; The networked communication module accesses the Internet of Things, realizes data transmission of the access control device and the data centralized device, and uploads the collected human body image features to the data centralized device.

Preferably, the video monitoring device specifically includes: a person target recognition module, a person target feature extraction module, a target trajectory analysis module, and an internet of things communication module; the person target recognition module is configured to extract a human body image region from each frame image And identifying a continuous human body image area in each frame picture as the same personal object target; the character target feature extraction module is configured to obtain a dress color feature of the person object, and select an identifiable face picture area from each frame picture. a video frame, extracting facial feature information of the person object, thereby obtaining a human body image feature of the character object; the target trajectory analysis module is configured to continuously change the position of the same personal object object recognized by each frame image, and extract the movement of the human target object Tracking, and confirming the associated room number of the person target according to the starting position or the ending position of the moving track; the IoT communication module is used to access the Internet of Things, and the collected human body image features, moving track information, and associated room numbers Upload to data set Set.

Preferably, the facial feature information is histogram data of each sub-region image texture feature value of the face region.

Further preferably, the facial feature information is obtained by decomposing a circumscribed rectangle of the entire face region into N×N sub-regions; for each of the sub-regions, extracting the pixel for each pixel in the sub-region a central pixel, a 3×3 pixel block including adjacent pixels of the upper left, upper, upper right, right, lower right, lower, lower left, and left sides of the pixel; the image texture feature value T _c of the central pixel is:

Where i _c represents the pixel gray value of the center pixel, and i _p represents the pixel gray value of the adjacent pixel. In the order of upper left, upper, upper right, right, lower right, lower, lower left, and left, the values of p are sequentially 1 to 8; and

.

Preferably, the dressing color feature is obtained by: calculating a color interval distribution histogram of the body region; and obtaining a plurality of color intervals in which the number of distributed pixels is the main color component according to the color interval distribution histogram, and the main color is The composition is used as the color of the body area.

Preferably, the body morphological feature is obtained by: obtaining a circumscribed rectangle of the human body image region, and acquiring a center point thereof according to the circumscribed rectangle; extracting an edge pixel of the human body image region; establishing the center point as a starting point, Each edge pixel is a feature vector of the end point, and all feature vector sets of the human body image area are feature vector groups as body shape features.

Preferably, the data concentrating device comprises: an Internet of Things base unit, a data concentrator unit, a data storage unit, a data association unit and a remote communication unit; the IoT base unit is used to cover the Internet of Things within the scope of the residential unit a wireless signal; the data concentrator unit receives the human body image feature uploaded by the access control device through the Internet of Things, the human body image feature uploaded by the video monitoring device, the movement track information, and the associated room number, and the Internet of Things instrument interface device uploads The data storage unit is configured to store a human body image feature uploaded by the access control device, a human body image feature uploaded by the video monitoring device, a movement track information, and an associated room number, and the Internet of Things instrument interface device uploads The meter reading unit is configured to identify a human body image feature uploaded by the access control device, a human body image feature uploaded by the video monitoring device, a movement track information, and an associated room number, and the Internet of Things instrument interface device uploads Meter reading Correlation, integration and storage of information associated with the data registered in the memory unit; said remote communication unit to transmit the association information to the remote registration data aggregation analysis large internet.

Preferably, the data associating unit compares the facial feature information of the entry and exit unit door personnel extracted by the access control device, the dress color feature with the facial feature information of the person target extracted by the video monitoring device, and the dress color feature. Determining whether the two correspond to the same personal object; in the case of confirming that the same personal object is confirmed, the data association unit assigns a human object identification number to the human object, and registers facial feature information for the human object identification number; and, The data association unit registers the person object entry and exit unit door time record and the person object unit according to the unit door entry and exit time of the person object registered by the access control device, and according to the movement track information of the person object uploaded by the video monitoring device and the associated room number. The moving time and the mobile floor record, the person object associated room number, and the time record of entering and leaving the associated room, thereby integrating the information related registration of the person object.

Preferably, the big data convergence analysis platform comprises: a personnel big data warehouse and a data analysis server; the personnel big data warehouse is used for registering the information association of each unit of each residential house in the jurisdiction uploaded by each data centralized device. Storage; the data analysis server is used to implement big data analysis related to personnel management.

(3) Beneficial effects

Compared with the prior art, the present invention provides a residential access control system that implements human data collection and analysis. The invention takes the personnel entering and leaving the residential cell unit as the object, and uses the access control device and the video monitoring device deployed by the cell unit to perform the image capturing of the personnel, thereby performing the extraction and recognition of the face, the dress and the physical features of the person, and acquiring the feature information and the association of the personnel. Information and behavioral model information; the system aggregates the above information into the big data platform, realizes big data analysis of personnel information, establishes portraits of personnel behaviors, identifies residential rooms, and integrates social management and public institutions of government agencies and community properties. Service support. The invention can comprehensively acquire and identify various types of personnel information, in particular to mine potential valuable information for personnel management, and realize diversified big data analysis; the invention can reduce the influence on the normal life rhythm of the residents, by means of deployment IoT equipment in residential communities, potentially completing the collection, identification and analysis of personnel information.

DRAWINGS

1 is a schematic view showing the overall structure of a residential access control system of the present invention;

2 is a schematic structural view of an access control device of a residential access control system of the present invention;

3 is a schematic structural view of a video monitoring device of a residential access control system of the present invention;

4 is a schematic structural view of a data concentrating device of a residential access control system of the present invention;

FIG. 5 is a schematic structural diagram of a big data convergence analysis platform of a residential access control system according to the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. It is obvious that the described embodiments are only a part of the embodiments of the present invention, but not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

FIG. 1 is a residential access control system for realizing large data collection of personnel provided by the present invention. The system includes: an access control device 101 installed in each unit door of the residential house, a video monitoring device 102 installed in each floor corridor of the residential house and covering the public space of the floor, and an Internet of Things located on each floor or each unit of the residential building. The meter interface device 103, the data concentrating device 104 installed in the communication room of each unit of the home, and the big data convergence analysis platform 105.

The access control device 101 is a smart device having a door lock control, a video intercom, a surveillance shooting, and an Internet of Things communication function. As shown in FIG. 2, the access control device 101 includes an electronic door lock 101A, a digital button 101B, a card reader or fingerprint identifier 101C, a walkie-talkie 101D, a front camera 101E, a rear camera 101F, an Internet of Things communication module 101G, and a main control chip. 101H. The electronic door lock 101A can lock or open the unit door under the control of the signal of the main control chip 101H. The number button 101B is used to input a room number. According to the input room number, the main control chip 101H activates the intercom 101D, and the operator screen is photographed by the front camera 101E to realize visual voice intercommunication with the designated room. The digital button 101B can also be used to input a password, and the main control chip 101H realizes password unlocking by comparing with a preset password. The card reader or fingerprint identifier 101C can read the access card number held by the operator or scan the operator fingerprint, and the main control chip 101H realizes the card unlocking or the fingerprint unlocking by comparing the preset card number or the owner fingerprint information. The main control chip 101H can also identify the facial feature information of the current access control operator by using the operator screen captured by the front camera 101E, and compare with the preset owner facial feature information to realize facial identity recognition unlocking. In this way, the owner who has registered the facial features can input the home room number using the numeric keypad 101B, and then face the front camera 101E to verify the face, thereby unlocking. The front camera 101E of the access control device faces the unit door and is used for capturing a monitoring video of a certain space outside the unit door, from which an access control operator and a person entering the unit door can be obtained. The rear camera 101F is oriented in the unit door, and a monitoring video of a certain spatial range in the unit door is taken, from which a screen of the person leaving the unit door can be obtained. The main control chip 101H obtains the picture acquired by the front camera 101E and the rear camera 101F; extracts the facial feature information of the access operator from the current operator picture taken by the front camera 101E, and uses the face identification unlocking; The human body image features of the person entering and leaving the unit door are extracted from the screens collected by the camera 101E and the rear camera 101F, and the entry and exit time of the person is recorded according to the shooting time. The Internet of Things communication module 101G of the access control device accesses the Internet of Things by using a communication protocol such as ZigBee, NB-IoT, GPRS, WI-FI, etc., and realizes data transmission between the access control device and the unit data concentrating device 104. The main control chip 101H uploads the human body image feature to the data concentrating device 104 through the Internet of Things through the Internet of Things communication module 101G.

The human body image features extracted by the main control chip 101H include facial feature information, dress color features, and body shape features of the person entering and leaving the cell door. For the screen of the person entering and leaving the unit door photographed by the front camera 101E and the rear camera 101F, the main control chip 101H first recognizes the human figure area existing therein, and further decomposes the body image area into the face area and the body area. The recognition of the human body image area can be realized by the contour extraction and the classification machine identification method based on the motion area detection; that is, the motion area is obtained by using the front and rear frame picture difference calculations captured by the camera; and the motion area contour is extracted by using a gradient judgment algorithm or the like; The vector group is established with the center point of the circumscribing rectangle of the contour of the motion region as the starting point and the edge pixel of the contour of the motion region as the end point; the vector group is brought into the SVM classifier to identify whether the motion region is a human body image region; The body image area is divided into a face area and a body area.

The facial feature information extracted by the main control chip 101H from the face region is the facial region image texture feature value, and the texture feature value is insensitive to the illumination change, so that the actual condition of the difference in the ambient illumination variation of the access control device can be well adapted; the texture feature value is The face image shooting angle offset also has a relatively strong adaptability, so that the person passing through the unit door does not affect the recognition even if the face is tilted to some extent relative to the access control device. The method for calculating the texture feature value of the facial region image is as follows:

Decomposing a circumscribed rectangle of the entire face region into N×N sub-regions, and the value of N ranges from 5 to 20; for each of the sub-regions, extracting the pixel as the center pixel for each pixel in the sub-region, including a 3×3 pixel block of adjacent pixels of the upper left, upper, upper right, right, lower right, lower, lower left, and left sides of the pixel; the image texture feature value T _c of the central pixel is:

That is to say, in the 3×3 pixel block, the gray value of the central pixel is the threshold value, and the gray value of the adjacent 8 pixels is compared with the grayscale value of the adjacent pixel, if the gray value of the adjacent pixel is greater than or equal to the central pixel grayscale. The value of the adjacent pixel is marked as 1, otherwise the adjacent pixel is marked as 0. In this way, 8 adjacent pixels in a 3×3 pixel block can be compared to generate 8 markers with a value of 0 or 1, and are adjacent in the order of upper left, upper, upper right, right, lower right, lower, lower left, and left. The mark corresponding to the pixel is arranged as an 8-bit binary number, and the 8-bit binary number is converted into decimal, that is, T _c , as the image texture feature value of the center pixel, and this value is used to reflect the texture information of the pixel block. For each of the N×N sub-regions, an image texture feature value of each pixel is obtained, and then histogram statistics of the sub-region pixel image texture feature values are performed, and histogram data of each sub-region is obtained; The histogram data of the sub-regions are combined, and the formed data set is used as facial feature information of the face region.

Further, the main control chip 101H counts the histogram of the color interval distribution of the body region, that is, the color value of each pixel of the body region is classified into one color interval in the color gamut, and the number of pixels distributed in each color interval is counted to form a color interval distribution. Histogram; according to the color interval distribution histogram, obtain the L color intervals with the largest number of distributed pixels, and take these color intervals as the main color components, where L is 10-20; the main color component is used as the dress of the body region Color characteristics.

The main control chip 101H also obtains a circumscribed rectangle of the human body image area, and obtains a center point thereof according to the circumscribed rectangle; further, the main control chip 101H extracts an edge pixel of the human body image area by using an edge recognition algorithm such as gradient judgment; As the starting point and the feature vector with each edge pixel as the end point, all the feature vector sets of the human body image region are the feature vector group as the body shape feature. The body morphological feature can be used for classification and identification of special morphological groups such as wheelchairs, limb disability, high months pregnant women, and young children.

The video monitoring device 102 of the system is used for realizing the security monitoring function on the one hand, and helps the property management department to detect abnormal situations such as fire and theft in time. On the other hand, the video monitoring device 102 is used to photograph a person monitoring screen indicating the activity of a person entering the home unit in the public space of the house before entering a certain room. Similarly, the video monitoring device 102 can extract the human body image features from the person monitoring screen, including facial feature information, dress color features, and body shape features of the person. Since the range of activities of the person in the public area of the house is large and subject to the shooting position, it is difficult for the video monitoring device 102 to always obtain a larger and clearer facial image of the person like the access control device 101, and with the personnel being far and near. Or by near and far movement, the size, position and shape of the human body image area in each frame are constantly changing. In many frame pictures, the face picture area of the person is unrecognizable, but in the temporally adjacent frames. The body image areas of the same personal object are in continuity with each other in position. In addition, although the facial feature information and body shape characteristics of the same person in each frame are relatively changed, the color characteristics of the dressing are kept stable. As shown in FIG. 3, the video monitoring device 102 specifically includes a person target recognition module 102A, a person target feature extraction module 102B, a target trajectory analysis module 102C, and an Internet of Things communication module 102D. The person target recognition module 102A is configured to extract a human body image region from each frame image, and identify a continuous human body image region in each frame image as the same personal object target; and according to the position of the center point of the human body image region of each frame image, The human figure area in which the center point position distance in the adjacent two frames is within a predetermined threshold is recognized as the same personal object target. The character target feature extraction module 102B is configured to obtain a dress color feature of the person object, and select a video frame having an identifiable face image region from each frame image to extract facial feature information of the person target; specifically, the character target The feature extraction module 102B extracts the dress color feature by statistically plotting the histogram of the color interval distribution according to the human body image region of the same personal object target in each frame image; and further, the person target feature extraction module 102B is based on the same personal object target in each frame image. The body image area determines the size of the face area. When the size of the face area meets a predetermined criterion, it is determined to belong to the identifiable face picture area, and facial feature information of the person object is extracted from the area. The target trajectory analysis module 102C continuously changes the position of the same personal object recognized by each frame picture, extracts the movement trajectory of the person object, and confirms the room number corresponding to the person target according to the start position or the end position of the movement trajectory, thereby The character target is associated with the corresponding room number. The Internet of Things communication module 102D accesses the Internet of Things, and uploads the human body image feature, the movement track information, and the associated room number of the person object collected by the video monitoring device 102 to the data concentrating device 104 through the Internet of Things.

The Internet of Things instrument interface device 103 connects the smart meter of the residential unit, such as a water, electricity and gas meter with communication functions, periodically collects the readings of the smart meter according to a fixed time interval; and the device accesses the Internet of Things, the collected The meter readings are uploaded to the data concentrating device 104 via the Internet of Things. The degree of change in meter readings over a certain time interval can indirectly reflect the living conditions of the personnel in the corresponding room.

The data concentrating device 104, as shown in FIG. 4, includes an Internet of Things base station unit 104A, a data concentrator unit 104B, a data memory unit 104C, a data association unit 104D, and a remote communication unit 104E. The Internet of Things base station unit 104A is configured to cover the Internet of Things wireless signal within the scope of the present residential unit, so that the access control device 101, the video monitoring device 102, and the Internet of Things instrument interface device 103 can access the Internet of Things, thereby uploading data to the data. The centralized device 104; the Internet of Things base station unit 104A can cover the unit based on communication protocols such as ZigBee, NB-IoT, GPRS, WI-FI, and allow and carry the unit access control device 101, the video monitoring device 102, and the Internet of Things instrument interface device 103. Access. The data concentrator unit 104B receives the information uploaded by the access control device 101, the video monitoring device 102, and the Internet of Things instrument interface device 103 through the Internet of Things, including the human body image features uploaded by the access control device 101, and the human body image features and movement trajectories uploaded by the video monitoring device 102. The information and the associated room number, meter readings uploaded by the Internet of Things meter interface device 103, and stored to the data memory unit 104C. The data association unit 104D is configured to identify the human body image features uploaded by the access control device 101, the human body image features uploaded by the video monitoring device 102, the movement track information, and the associated room number, and the correlation between the meter readings uploaded by the Internet of Things instrument interface device 103. The information association registration is integrated for the above information, and the information relation registration data is stored in the data memory unit 104C. The remote communication unit 104E accesses the wide area network based on the 3G or 4G communication technology, and remotely transmits the information associativity registration stored by the data storage unit 104C to the big data convergence analysis platform 105.

Specifically, the human body image feature uploaded by the data association unit 104D to the access control device 101, the human body image feature uploaded by the video monitoring device 102, the movement track information, and the associated room number, and the correlation identification of the meter readings uploaded by the Internet of Things instrument interface device 103 are specifically described. process. As described above, the access control device 101 extracts the facial feature information and the dress color feature of each entry and exit unit door person; the human body image features uploaded by the video monitoring device 102 include the dress color feature and the face feature information of the person target; data association The unit 104D compares the facial feature information of the entry and exit unit door personnel, the dress color feature with the facial feature information of the person target, and the dress color feature to determine whether the two correspond to the same personal object object; the specific process of the comparison is: The entry and exit unit door facial feature information is represented as F _i,j , where i=1, 2, 3, ..., N×N represents N×N sub-regions in which the face region is decomposed, and j represents the statistical value of the sub-region. Each column in the image texture feature value histogram, F _i,j represents the value of the jth column of the i-th sub-region histogram; similarly, the facial feature information of the person object extracted by the video monitoring device is represented as M _i,j ,M _{i, j} represents the value of the jth column of the i-th sub-region histogram; then the facial feature information difference degree S is expressed as

Where α _i represents the weight value of each of the N×N sub-regions, and the sub-region weight values at key positions such as eyes and mouth are larger. If the calculated degree of difference S is smaller than the first threshold, it is confirmed that the person entering and leaving the unit door belongs to the same person object as the person object extracted by the video monitoring device. If the degree of difference S is greater than or equal to the first threshold but less than the second threshold, the color characteristics of the person in the entrance and exit unit and the person target extracted by the video monitoring device are obtained, and whether the main color components of the two are coincident, and if they overlap, the entry and exit unit is confirmed. The person selected by the door and the video surveillance device belong to the same personal object. If the degree of difference S is greater than the second threshold, or S is greater than or equal to the first threshold and less than the second threshold, but the main color components do not coincide, it is determined that the person entering and leaving the unit is not the same person object as the person object extracted by the video monitoring device . In the case of confirming that the same personal object is confirmed, the data associating unit 104D assigns a human object identification number to the human object, and registers its facial feature information for the human object identification number; and the data associating unit 104D registers according to the access control device 101. The unit door entry and exit time of the person object is registered according to the movement track information of the person object uploaded by the video monitoring device 102 and the associated room number, and the person object entry and exit unit door time record, the movement time in the person object unit, and the moving floor are registered. The record, the person object associated room number, and the time record of entering and leaving the associated room, thereby integrating the information related registration of the person object. The data association unit 104D further extracts a feature vector group of the body shape feature according to the access control device 101, and substitutes the feature vector group into the trained SVM classifier to identify whether it belongs to a special group such as a wheelchair, a physical disability, a high month pregnant woman, and a child; If it is recognized that the person object belongs to a special group, a special crowd type flag is recorded in the information related registration of the person object. In addition, for each recognized person object, firstly, the facial feature information is compared with the facial feature information of the already registered person object, and it is determined whether the currently recognized person object has the information relevance registration; if there is no information relevance Registration, a new information association registration is created for the currently identified person object; conversely, if the previously identified person object already has information association registration, the current entry and exit unit door time record and the unit are recorded in the registration. Move time and mobile floor record, associated room number, time record of entering and leaving the associated room. The data association unit 104D also integrates information association registration for each room, including the person object identification number associated with the room, the time record of each character object entering and exiting the associated room, and the meter reading record for the room.

The big data convergence analysis platform 105 is a cloud data center responsible for big data aggregation and analysis of residential personnel within a certain jurisdiction. The big data convergence analysis platform 105 receives the information relevance registration data uploaded by all the data concentrating devices 104 within the jurisdiction area through the wide area network. As shown in FIG. 5, the big data convergence analysis platform 105 includes a human big data warehouse 105A and a data analysis server 105B. The person big data warehouse 105A is used to store the information related registration data of each unit of each house in the area uploaded by each data concentrating device 104, so that the person big data warehouse 105A gathers a large number of personnel information records, including characters. Object entry and exit unit door time record, activity record of person object unit in unit, record of person object associated room number, time record of person object entering and leaving associated room, special crowd mark of person object; also including each room of jurisdiction residential area Person object records, personnel entry and exit time records, meter readings. The data analysis server 105B realizes big data analysis related to personnel management based on the mass personnel information record stored in the person big data warehouse 105A. Specifically, the data analysis server 105B can perform the establishment and analysis of the character object behavior portrait, according to each registered person object, according to the entry and exit unit door time record, the unit activity record, the associated room number record, In and out of the associated room time record and special crowd mark, statistics on the distribution of these records, extract the character object in and out of the unit, the time period in which the inbound and outbound rooms are mainly distributed, the length of time and the length of the character object in the public area of the unit, the character Dimensional features such as the number of rooms associated with the object, and the behavior of the character object is established. According to the character object behavior portrait, different types of people can be effectively distinguished. For example, if a character object behavior image reflects that the task object is active in the public area of the residential unit and the track length is significantly longer than the normal distribution interval, the character The subject is probably not an ordinary residential resident, but may be someone who is posting a distribution advertisement in a public area of the home. The data analysis server 105B can also perform statistical analysis of indicators for each room, count the number of person objects associated with each room, the time distribution of the person objects entering and leaving the room, and the amount of water and electricity, and then perform abnormal monitoring according to the statistical analysis results of the indicators; for example, if The number of person objects associated with a room is significantly higher than the normal number, or the time distribution of the character object entering and leaving the room is highly deviated from the regular time distribution, or the water and electricity usage is significantly higher than the normal range, then the room may belong to the group rent house. Or the house is privately operated for commercial operation, and the relevant management department may conduct necessary inspection visits to the house. The data analysis server 105B can also classify the person objects with special tags and the corresponding associated room numbers to give personalized public or property services, such as giving priority assistance in the event of an accident.

The relevant modules involved in the system are hardware system modules or functional modules combined with computer software programs or protocols and hardware in the prior art, and the computer software programs or protocols involved in the functional modules are all in the field. A technique known to the skilled person is not an improvement of the system; the improvement of the system is an interaction relationship or a connection relationship between the modules, that is, an improvement of the overall structure of the system to solve the problem to be solved by the system. Corresponding technical issues.

While the embodiments of the present invention have been shown and described, it will be understood by those skilled in the art The scope of the invention is defined by the appended claims and their equivalents.

Claims

A residential access control system for realizing human data collection and analysis, comprising: an access control device, a video monitoring device, an Internet of Things instrument interface device, a data centralized device, and a big data convergence analysis platform;

The access control device is installed in each unit door of the residential building, and is used for door lock control, visual intercom, and monitoring the shooting of the person entering and leaving the unit door; extracting the human body image of the person entering and leaving the unit door from the captured picture a feature, the human body image feature includes facial feature information, a dress color feature, a body shape feature, and records the entry and exit time of the person according to the shooting time; and uploads the human body image feature to the data through the Internet of Things through the Internet of Things communication Concentrated device

The video monitoring device is configured to capture a person monitoring screen of a person entering the residential unit in the public space of the house, extracting a person object from the person monitoring screen, obtaining a human body image feature, a moving track information, and an associated room number of the person target, The Internet of Things is uploaded to the data concentrator;

The IoT instrument interface device is connected to the smart meter of the residential unit, and is used for periodically collecting the readings of the smart meter according to the fixed time interval, and uploading the collected meter readings to the data centralized device through the Internet of Things;

The data concentrating device is configured to cover the Internet of Things within the scope of the residential unit, and receive and store the human body image feature uploaded by the access control device, the human body image feature uploaded by the video monitoring device, the movement track information, and the associated room number through the Internet of Things. And reading the meter reading uploaded by the Internet of Things instrument interface device; identifying the human body image feature uploaded by the access control device, the human body image feature uploaded by the video monitoring device, the moving track information, the associated room number, and the Internet of Things instrument interface Correlation between meter readings uploaded by the device, integrating and storing information association registration; accessing the wide area network and transmitting the information association registration remotely to the big data convergence analysis platform;

The big data convergence analysis platform is configured to receive and store the information related registration uploaded by all data concentrating devices in the jurisdiction area through the wide area network, and apply the information related registration to perform big data analysis related to personnel management.
The residential access control system for implementing human data collection and analysis according to claim 1, wherein the access control device comprises: an electronic door lock, a digital button, a card reader or a fingerprint recognizer, a walkie-talkie, a front camera, and a rear The camera, the Internet of Things communication module and the main control chip; the electronic door lock can be locked or opened under the control of the main control chip; the digital button is used to input the room number and password; the main control chip starts the walkie-talkie according to the input room number And the operator screen is photographed by the front camera to realize visual voice intercommunication with the designated room; the main control chip also compares the input password with the preset password to realize password unlocking; the card reader or the fingerprint recognizer can Reading the access card number or scanning the operator's fingerprint held by the operator, the main control chip realizes the card unlocking or fingerprint unlocking by comparing the read in/out card number or the scanned operator fingerprint with the preset card number or the owner fingerprint information. The main control chip uses the operator's picture taken by the front camera to identify the current access control The author's facial feature information is compared with the preset owner's facial feature information to realize facial identity recognition unlocking; the front camera is used to capture a surveillance video of a certain space outside the unit door, from which the access control operator and the personnel entering the unit door are obtained. The rear camera is used to capture the surveillance video of a certain space within the unit door, and obtains the picture of the person leaving the unit door; the main control chip also extracts the personnel entering and leaving the unit door from the pictures taken by the front camera and the rear camera. The human body image features, and records the entry and exit time of the person according to the shooting time; the Internet of Things communication module accesses the Internet of Things, realizes data transmission of the access control device and the data centralized device, and uploads the collected human body image features to the data centralized device.
The residential access control system for implementing human data collection and analysis according to claim 1, wherein the video monitoring device comprises: a person target recognition module, a character target feature extraction module, a target trajectory analysis module, and an Internet of Things communication module. The character target recognition module is configured to extract a human body image region from each frame image, and identify a continuous human body image region in each frame image as a same personal object target; the character target feature extraction module is configured to obtain the character target Dressing a color feature, and selecting a video frame having an identifiable face picture area from each frame picture, extracting facial feature information of the person object, thereby obtaining a human body image feature of the person object; the target trajectory analysis module is configured to The identified position of the same personal object continuously changes, extracts the movement trajectory of the person target, and confirms the associated room number of the person target according to the starting position or the ending position of the moving trajectory; the IoT communication module is used for accessing Internet of Things, and will be collected The human body image features, the movement track information, and the associated room number are uploaded to the data concentrating device.
The residential access control system for implementing human data collection and analysis according to claim 1, wherein the facial feature information is histogram data of image texture feature values of each sub-region of the face region.
The residential access control system for implementing human data collection and analysis according to claim 4, wherein the facial feature information is obtained by decomposing a circumscribed rectangle of the entire face region into N×N sub-regions; a sub-region, for each pixel in the sub-region, extracting a 3×3 pixel block with the pixel as the center pixel, including the upper left, upper, upper right, right, lower right, lower, lower left, and left adjacent pixels of the pixel The image texture feature value T c of the center pixel is:

Where i c represents the pixel gray value of the center pixel, and i p represents the pixel gray value of the adjacent pixel. In the order of upper left, upper, upper right, right, lower right, lower, lower left, and left, the values of p are sequentially 1 to 8; and
The residential access control system for implementing human data collection and analysis according to claim 1, wherein the dress color feature is obtained by: calculating a color interval distribution histogram of the body region; and acquiring the histogram according to the color interval distribution. The plurality of color intervals in which the number of distributed pixels is the largest are used as the main color component, and the main color component is used as the color feature of the body region.
The residential access control system for realizing human data collection and analysis according to claim 1, wherein the body shape feature is obtained by: obtaining a circumscribed rectangle of the human body image region, and obtaining a center point thereof according to the circumscribed rectangle; Extracting the edge pixels of the human body image region; establishing a feature vector starting from the center point and ending with each edge pixel, and all feature vector sets of the human body image region are feature vector groups as body shape features.
The residential access control system for implementing human data collection and analysis according to claim 1, wherein the data concentrating device comprises: an Internet of Things base station unit, a data concentrator unit, a data storage unit, a data association unit, and a remote communication unit. The IoT base station unit is configured to cover an Internet of Things wireless signal within a range of the residential unit; the data concentrator unit receives the human body image feature uploaded by the access control device through the Internet of Things, and the human body image uploaded by the video monitoring device a feature, a movement track information, and an associated room number, a meter reading uploaded by the Internet of Things meter interface device; the data storage unit is configured to store a human body image uploaded by the access control device, and a human body image uploaded by the video monitoring device a feature, a movement track information, and an associated room number, a meter reading uploaded by the Internet of Things meter interface device; the data association unit is configured to identify a human body image feature uploaded by the access control device, and a human body image uploaded by the video monitoring device Feature, movement track information And correlating the room number, the meter readings uploaded by the Internet of Things meter interface device, integrating and storing the information association registration in the data storage unit; the remote communication unit registering the information association Remotely sent to the big data aggregation analysis platform.
The residential access control system for implementing human data collection and analysis according to claim 8, wherein the data associating unit extracts facial feature information, dress color characteristics and video by the entry and exit unit door personnel extracted by the access control device. The face feature information and the dress color feature of the person object extracted by the monitoring device are compared to determine whether the two correspond to the same personal object object; and in the case of confirming that the same personal object object is confirmed, the data associating unit assigns a person object to the person object An identification number, and registering facial feature information for the human object identification number; and the data association unit according to the unit door entry and exit time of the human object registered by the access control device, and according to the movement trajectory information of the human object uploaded by the video monitoring device and Associate the room number, register the person object entry and exit unit door time record, the movement time and moving floor record in the person object unit, the person object associated room number, and the time record of entering and leaving the associated room, thereby integrating the information related registration of the person object .
The residential access control system for implementing human data collection and analysis according to claim 1, wherein the big data convergence analysis platform comprises: a personnel big data warehouse and a data analysis server; and a human big data warehouse is used for each data. The information association registration of each unit of each residential area in the jurisdiction uploaded by the centralized device is stored; the data analysis server is used to implement big data analysis related to personnel management.