CN113360241A - Traffic tunnel analysis platform - Google Patents

Abstract

The application belongs to the technical field of tunnel inspection, and particularly relates to a traffic tunnel analysis platform. The method comprises the following steps: the basic hardware layer adopts a cluster server architecture; the system comprises a basic cloud platform, a HASEN container, a cloud service platform and a communication platform, wherein the HASEN container is adopted to provide a lightweight distributed resource management and scheduling system, and a management system based on clouding, servitization, opening unification and telecommunication level is constructed; the video cloud component comprises a video management cloud component and a video analysis cloud component, the video management cloud component is used for managing a monitoring process and providing processing for real-time monitoring data, and the video analysis cloud component is used for identifying the acquired images and analyzing personnel and personnel behaviors. The traffic tunnel analysis platform has the advantages of light weight, cloudization, convenience in deployment, convenience in management and the like.

Description

Traffic tunnel analysis platform

Technical Field

The application belongs to the technical field of tunnel inspection, and particularly relates to a traffic tunnel analysis platform.

Background

In a traditional monitoring system, an IPC video is stored in planned fixed storage equipment, the storage capacity is changed due to the fact that the storage time of the IPC video is increased or reduced, when the number of the storage equipment is changed, the IPC of the whole system needs to be re-planned, the IPC video is assigned to the corresponding storage equipment manually, generally, an industry N +1 cluster only has a backup function, the IPC loaded on the cluster is a certain member of a fixed loading cluster, and dynamic adjustment cannot be achieved.

Disclosure of Invention

In order to solve at least one of the above technical problems, the present application provides a traffic tunnel analysis platform to implement resource integration and unified management.

The application provides a traffic tunnel analysis method, which comprises the following steps:

the basic hardware layer adopts a cluster server architecture;

the system comprises a basic cloud platform, a HASEN container, a cloud service platform and a communication platform, wherein the HASEN container is adopted to provide a lightweight distributed resource management and scheduling system, and a management system based on clouding, servitization, opening unification and telecommunication level is constructed;

the video cloud component comprises a video management cloud component and a video analysis cloud component, the video management cloud component is used for managing a monitoring process and providing processing for real-time monitoring data, and the video analysis cloud component is used for identifying the acquired images and analyzing personnel and personnel behaviors.

Preferably, the cluster server architecture includes 4U dual or single storage rack servers, and 2U2 rack servers.

Preferably, the application lifecycle management module comprises:

the IPC adding unit is used for loading the newly added IPC to the cloud node with lower load according to the load of the current cluster member;

the cloud node regulating and controlling unit is used for newly increasing or reducing cloud node members;

and the IPC adjusting unit is used for sharing the IPC load on the fault cloud node to other cloud nodes and returning the migrated IPC load to the original cloud node after the fault cloud node is repaired.

Preferably, the video management cloud component includes:

the video system management unit is used for responding to the user request and setting and changing the communication parameters, the monitoring parameters and the authority parameters of the video equipment;

the real-time monitoring unit is used for responding to a user request and setting and changing the style parameters of the video monitoring terminal;

the video management unit is used for responding to a configuration request of a user for video recording of the camera, recording the video and processing video data;

the television wall control unit is used for controlling decoding display of images on the plurality of large-screen liquid crystal display units;

and the front end integrated unit is used for detecting the invasion and the arrangement of personnel and vehicle tripwires in the video and sending out an alarm signal.

Preferably, the video system management unit includes:

the equipment management unit is used for providing access and disconnection to the front-end equipment;

the communication management unit is used for connecting external domain equipment and providing an external domain information configuration port for a user with management authority;

the front-end equipment monitoring parameter adjusting unit is used for setting and changing the parameters of a preset position, a guard position and a cruise track of the front-end video equipment;

and the user parameter management unit is used for managing roles, users and authorities.

Preferably, in the real-time monitoring unit, the style parameters of the video monitoring end include window layout, split screen, and display scale.

Preferably, the video parsing cloud component includes:

the face recognition unit is used for extracting the features of face data in the video or the picture to form a face feature database so as to retrieve the face data in the face database through the face picture;

the human body image searching unit is used for searching the video information through the human body image so as to locate the target human body information;

and the behavior analysis unit is used for carrying out perimeter intrusion detection in the monitoring area and giving an alarm to the intrusion behavior when the intrusion row is detected in the set area.

Preferably, the face recognition unit includes:

the human face real-time control unit is used for capturing a human face in real time through the front-end camera, identifying the captured human face and storing human face information into the human face feature resource library;

the face dynamic library retrieval unit is used for traversing and comparing the uploaded face data with face feature data in a face snapshot library and displaying a comparison result;

the face static library retrieval unit comprises a witness comparison unit and a static library retrieval unit, wherein the witness comparison unit is used for calculating the similarity according to two input face pictures, and the static library retrieval unit is used for returning the face pictures with the similarity above a threshold value according to one input face picture.

The method and the system change various defects of traditional chimney type construction, such as information isolated island, difficulty in efficient utilization of data, difficulty in butt joint among platforms and the like. By regularly collecting the states and load conditions of the members of each video monitoring cloud node, the system can be adjusted in time when the load is seriously unbalanced. The metadata security is guaranteed, and the overall performance of the system is guaranteed to be optimal on the basis of the data security.

Drawings

Fig. 1 is a diagram of a traffic tunnel analysis platform architecture according to the present application.

Detailed Description

In order to make the implementation objects, technical solutions and advantages of the present application clearer, the technical solutions in the embodiments of the present application will be described in more detail below with reference to the accompanying drawings in the embodiments of the present application. In the drawings, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The described embodiments are some, but not all embodiments of the present application. The embodiments described below with reference to the drawings are exemplary and intended to be used for explaining the present application, and should not be construed as limiting the present application. All other embodiments obtained by a person of ordinary skill in the art without any inventive work based on the embodiments in the present application are within the scope of protection of the present application. Embodiments of the present application will be described in detail below with reference to the drawings.

The machine room deployment lightweight cloud platform mainly achieves pooling management of physical equipment such as computing and storage through a clouding technology, achieves capabilities of software defined infrastructure (namely software defined computing and software defined storage), provides an open, quick and efficient clouding resource environment, and supports various video cloud services in a resource serving mode. The basic cloud platform realizes unified management and video retrieval of the front-end camera. The analysis cloud platform realizes intelligent analysis of videos, and the intelligent analysis comprises video abstraction, video retrieval, face recognition, human body behavior analysis and the like.

Therefore, the application provides a traffic tunnel analysis platform, which mainly comprises a basic hardware layer, a basic cloud platform, a video cloud component and an open service layer as shown in fig. 1. The method comprises the following specific steps:

the basic hardware layer adopts a cluster server architecture, and in some optional embodiments, the basic hardware layer consists of a brand new generation of 4U double-path or single-path storage rack server with low energy consumption, strong expansion capability, high reliability, easy management and easy deployment and a 2U 2-path rack server with wide users.

The HASEN container is adopted to provide a lightweight distributed resource management scheduling system, a management system based on clouding, servitization, opening unification and telecommunication level is constructed, the management system comprises an application life cycle management module and an application daily operation and maintenance management module, and the application life cycle management module is used for carrying out resource isolation and scheduling on the HASEN container.

In some alternative embodiments, the base cloud platform consists of HSASEN containers and CSP Edge software. The HASEN container technology is adopted, and a lightweight distributed resource management scheduling system is provided. The CSP (cloud Service platform) Edge is constructed based on the design concept of cloud, Service, open unification and telecom level, provides a unified and efficient Web-based visual operation maintenance function for upper-layer application, and mainly comprises two aspects of application life cycle management and application daily operation and maintenance. The application life cycle management comprises container resource isolation, scheduling, application deployment, service management, elastic expansion and gray scale upgrading.

In the application, the HASEN container technology adopted by the basic cloud platform is a lightweight distributed resource management scheduling system. The method has the following characteristics:

(1) light weight: the small-scale cluster is efficiently managed, the resource occupation on the computing node is less than 100MB, the distributed selection master and the distributed cache are embedded, a third party is not required to be relied on, and the performance is stronger; the light communication bus optimizes communication performance.

(2) The deployment is convenient: compared with a virtualization technology, the container technology improves the utilization efficiency of memory, storage and bandwidth resources; in the running process, because the volume is small, the container can be started in a second level generally, the virtual machine needs to call a large amount of resources, and the starting time is about 20 minutes.

(3) The management is convenient: the container technology can complete the management of a single container through a simple command line, and complete the quick packing and migration of the mirror image; meanwhile, the management of large-scale container clusters can be realized through the arrangement tool.

(4) Management of large/small clusters is supported while maintaining scheduling performance without degradation.

(5) And shielding the difference between different hardware and the bottom operating environment.

(6) Cluster automation treatment: and the cluster is automatically established without complex configuration.

(7) Dynamically deploying fine-grained application in a process: and fine-grained deployment of functions, dynamic libraries and the like is supported.

(8) Intelligent scheduling and high resource utilization: and a plurality of resource scheduling strategies are provided, resource overspending, priority preemption and dynamic application resource adjustment are supported, and the overall resource utilization rate is improved on the premise of meeting application multi-dimensional resource constraints.

(9) Sharing cluster resources: supporting multi-frame shared cluster resources (Spark, Hadoop, etc.); supporting scheduling of different traffic types (long service, short tasks, timed tasks).

(10) Opening property: support the mainstream framework of the docking industry, such as meso, k8s, spark, hadoop and the like.

(11) High reliability: and the automatic restart of the failed task is supported, and the service is not interrupted.

(12) The realization of the micro-service architecture is facilitated: second-level creation and execution of containers facilitates construction of microservices; the orchestration tool of containers facilitates fast assembly and scheduling of microservices.

(13) Elastic expansion: because the container units are mutually independent and are managed by a uniform arrangement tool, and the arrangement tool has the function of finding container nodes, the elastic expansion of the container can be automatically completed in a short time; meanwhile, each container is an independent individual, and the use of the resources called by the container and the use of the container are managed by the arranging tool, so that the use of the whole container system is not influenced by reducing a certain container node.

(14) High availability: similar to elastic expansion, when a certain container node fails, the container arrangement tool can find the change of the node in time and make adjustment in time according to the external request condition, the use of the whole container system is not influenced, and the high availability of the system is realized.

The cloud cluster is a cluster formed by a plurality of cloud nodes (MPU, media processing unit) in a domain and operates in a cluster mode. When the IPC is added, the cluster management cloud node adds the newly added IPC to the appropriate cloud node according to the load of the current cluster member. The cluster can be dynamically expanded and contracted and comprises a newly added member and a deleted member; when the load of a certain cloud node is overlarge, supporting the sharing of partial services on the cloud node equipment to other cloud nodes; when a certain cloud node fails, the method supports the sharing of the service load on the failed equipment to other cloud nodes; when a certain cloud node is recovered from a fault, partial services on other cloud node equipment are supported to be shared to the cloud node. To summarize, the application lifecycle management module comprises:

2.1IPC adding unit, used for loading the newly added IPC to the cloud node with lower load according to the load of the current cluster member, along with the scale of the video monitoring system becoming larger and larger, the front end IPC quantity continuously increases, the video monitoring system needs to deal with the storage of mass data, at the initial stage of system construction, it needs to plan which cloud node each IPC needs to be deployed to, and the workload is huge. When the IPC is added, the cluster management service loads the newly added IPC to a proper cloud node according to the load of the current cluster member.

2.2 cloud node regulation and control unit, for adding or reducing cloud node members, the cloud cluster can support the members to be dynamically retractable, the support system can dynamically add or reduce cloud node members, the IPC is dynamically distributed to the appropriate cloud nodes by the cluster management cloud nodes, and manual operation is not needed.

And the IPC adjusting unit is used for sharing the IPC load on the fault cloud node to other cloud nodes and transferring the transferred IPC load back to the original cloud node after the fault cloud node is repaired. In the cloud cluster support extreme condition, when a plurality of cloud nodes are failed at the same time, IPC load on the failed cloud node is shared to other cloud nodes, and after the cloud node failure is recovered, partial service on other cloud node equipment is shared to the cloud node, so that the reliability is higher compared with that of an N +1 cluster.

By the mode, the cloud cluster supports dynamic adjustment of which cloud node the IPC is loaded to according to the load on each cloud node in the cluster, so that load balance of each cloud node in the cluster is guaranteed, members of the cluster are prevented from causing faults due to overlarge load of single equipment, and reliability of the system is improved.

And the video cloud component comprises a video management cloud component and a video analysis cloud component, the video management cloud component is used for managing the monitoring process and providing processing for real-time monitoring data, and the video analysis cloud component is used for identifying the acquired image and analyzing personnel and personnel behaviors.

The video management cloud component is an intelligent video monitoring product which is customized and developed based on the industrial video monitoring service requirements and combined with a video cloud storage product.

The video management cloud platform not only provides functions of real-time monitoring, cloud mirror control, video storage, electronic maps, television walls, alarm linkage and the like to meet basic monitoring requirements, but also provides advanced functions of multi-level multi-domain management, step-by-step forwarding, outer domain forwarding and the like.

The video management cloud component supports functions of domain main service dual-machine deployment, media service cluster deployment, video backup and the like, and ensures high reliability of monitoring service and data storage.

The platform system function is divided into five functional modules: the system management, the real-time monitoring function, the video recording management, the television wall function and the intelligent front-end management are integrated.

Specifically, the video management cloud component comprises:

and 3.1, the video system management unit is used for responding to the user request and setting and changing the communication parameters, the monitoring parameters and the authority parameters of the video equipment. The main work of the video system management unit includes:

equipment management: the video management cloud component provides functions of accessing and managing front-end equipment (IPC), and can uniformly manage all the access equipment. Management of external domain information: the video management cloud component provides an external domain information management function, and users with external domain management authority can configure external domain information. Managing a video recording plan: the video management cloud component provides a video plan management function, and a user can allocate, modify, delete and inquire a video plan for a video camera. Management of a cradle head preset position: the video management cloud component provides a function of managing the preset bits of the front-end video equipment, and a user can uniformly manage the preset bits of all accessed video equipment. And (3) guard management: the video management cloud component provides a function of managing the guard positions of the front-end video equipment, and a user can uniformly manage all the guard positions of the accessed video equipment. Cruise track management: the video management cloud component provides a function of managing a cruise track of the front-end video equipment, and the cruise track is composed of a plurality of preset fixed points. Alarm linkage management: the video management cloud component provides management for alarm linkage, and a user can uniformly manage all alarm linkages in the system. And (3) alarm log management: the video management cloud component provides management for the alarm logs, and a user can uniformly manage all the alarm logs in the system. Managing a system log: the video management cloud component provides a log information management function for the system running state and the user historical operation, and the user can know the running state of the system and the user operation record according to the log information. And (3) role management: the video management cloud component provides a management function for all roles, and the user can perform batch configuration on the authority of the user in the system through the roles. User management: the video management cloud component provides a user management function, and an administrator user (particularly a super administrator) can uniformly manage all users in the system. The system can manage all operation keyboards and users, and can set different authorities. Different authorities have different monitoring levels for different resources, and a system administrator can freely set the user level and the authority of the user for the equipment so as to facilitate management. Users can be divided into a number of different levels. The users in different levels can browse any one front-end image at the same time, but the control right is acquired according to the user level. The user with higher level has priority control right, and the user with lower level can continue to control after the user with higher level releases the control right; when the users in the same level preempt the same control right, the control right is obtained according to the principle of first-come first-serve. The control ability of the user to the system resource is only limited by the authority and priority, and is independent of the region where the resource is located.

And 3.2, the real-time monitoring unit is used for responding to the user request and setting and changing the style parameters of the video monitoring terminal.

The client is a main carrier for video monitoring service presentation and is a powerful tool for a client to use a monitoring system. And real-time video browsing of the monitoring points with the permission is checked on the PC through client software in real time. The main contents of the real-time monitoring unit comprise:

multiple layouts: the client supports the window layout of various split screen numbers: 1. 3, 4, 6, 8, 9, 10, 13, 14, 16, 25, 36, 49, 64, the number of split screens can be set using the split screen option button; double clicking the camera to start the video, selecting a playing window from left to right and from top to bottom, and if all the panes are used up, manually adding another multi-pane layout; the system client supports the simultaneous playing of real-time videos of a plurality of front-end devices. The maximum live preview may support 8 windows (in terms of 10804 Mbps 30 fps). Video direct connection: when an operator needs to check the optimal real-time dynamic state, the video front end can be directly connected, and the video real-time performance is improved. Adjusting the display proportion: the style of live display can be set, and the full coverage of the pane and the selection of the original proportion size are supported. Capturing pictures: when watching a real-time video or a video, a user can intercept a required picture. The user can grab the picture by a single picture or can grab the pictures by a plurality of pictures continuously. And the video playback in the real-time monitoring window is supported, so that the real-time video and the video can be compared conveniently. Fuzzy query: the fuzzy query function of the front-end monitoring point is supported, and the camera to be checked can be quickly searched through the keywords belonging to the name of the camera. Double-code stream video preview: the client can flexibly select the main code stream or the sub-code stream of the real-time preview front end, the network environment of the client is better, and when the bandwidth is higher, the main code stream with high code rate can be used to obtain a clearer picture; when the network environment of the client is poor and the bandwidth is low, the auxiliary code stream with low code rate can be used to obtain a smoother picture. Digital scaling: the digital zooming of the video which is monitored in real time and played back by the video is supported, the zooming function of the picture can be realized, and more picture details can be checked.

And 3.3, the video recording management unit is used for responding to a configuration request of a user for video recording of the camera, recording the video and processing the video recording data. The video management unit mainly comprises two video services of local video recording and platform video recording of the client. The platform video is divided into two video services of event video and timing video.

After the user starts real-time browsing at the client, the user can perform local video recording operation and store the video recording file on a local PC. The main contents of the local video include:

video file format: the video file format stored on the local PC is mp 4. Video retrieval: and the retrieval of the local video of the client is supported. The playing mode is as follows: the video file is played through the client side, and the video file is played by using a universal player.

The main contents of the platform video include:

(1) platform event recording

The platform event video belongs to a linkage mode of alarm linkage. When an alarm event occurs, the system starts the video recording task of the front-end equipment or other equipment with the alarm according to the linkage strategy, and stores the video recording file in the storage equipment, the video recording duration can be configured by a user, and the event video recording is stopped after the video recording duration is finished. Setting an event: the user can set the event triggering the video recording. Events that may be set include, but are not limited to: motion detection and switching value alarm.

(2) Platform timing video

The user can configure any video recording plan (including the start and end time of the video recording and the number of days of the video recording) of the video camera in the monitoring system, and the system carries out platform video recording at the specified time according to the video recording plans. The timed recording function is convenient for the user to continuously record all day and within a specific time period. Setting the video recording time: the user can set a timer recording strategy, and the recording start time, the recording end time and the recording days are specified in the strategy. Setting the front-end equipment of the timing video: the user may specify the front-end equipment for the timer recording.

The user can play back the video on the client, or download the system video file to the local PC (video file format is mp4), supporting the use of a universal player for playback. The method specifically comprises the following steps:

video recording label: and the method supports the marking of the label information during the playback of the video, and facilitates the subsequent retrieval. Video retrieval: the user can search the videos after the events, the videos of the events occurring before can be checked through the videos, the function of video monitoring evidence obtaining after the events is achieved, and the efficiency of searching the videos by the user can be improved according to the intelligent search of the events and the alarm. Recording and playing back: the user can play the post-event video, and the video of the pre-event scene can be checked through the video, so that the function of video monitoring and evidence obtaining after the event is realized. The method supports the simultaneous playing of videos of the same camera in different time periods in multiple panes, supports the setting of the number and layout of playing panes, automatically segments according to the number of panes, and can perform independent playback control on each playing window during segmented playback, wherein the maximum number of panes is 16. And (3) playback control: the method supports the acceleration of the playing speed of the video file, and can set the playing speed to be 2 times, 4 times, 8 times and 16 times of the normal speed; the slow video file playing speed is supported to be reduced, and the playing speed can be set to be 1/2 times and 1/4 times of the normal speed; the video file is played in a backing mode, and the playing speed can be set to be 2 times, 4 times, 8 times or 16 times fast backing playing; and single frame playing is supported, and only one frame of picture is played each time, so that a user can conveniently watch picture details. Synchronous playback: the method supports 9-path video synchronous playback, can simultaneously play back a plurality of video records, performs synchronous video playback control (synchronous fast forward, slow play, synchronous skip to a specified time point and the like), and is convenient for a user to compare monitoring videos in different places. Downloading the video: the user can store the platform video to the local part of the client for subsequent checking or publishing, and normal downloading and high-speed downloading are supported. Video recording label: and the method supports the marking of the label information during the playback of the live video and the video, and facilitates the subsequent retrieval. The method supports the presentation of video retrieval results in a time axis form, and can definitely identify whether the current time period has video, which is alarm video and which is plan video; the video at the appointed time point can be quickly positioned and played. Digital scaling: by digital zooming, the zooming function of the picture can be realized during playback of the video, and more picture details can be checked.

And 3.4, the video wall control unit is used for controlling the decoding display of the images on the plurality of large-screen liquid crystal display units.

The television wall is a monitoring device commonly used by a monitoring center, consists of a plurality of large-screen liquid crystal display units, can intensively display a plurality of front-end video pictures, is convenient for monitoring personnel to watch, and is supported by a system to upload real-time monitoring pictures to the television wall through a decoder for playing.

The television wall has the functions as follows: and the television wall decoder control module at the C/S software client side is supported to carry out layout configuration on the television wall, and the television wall and the decoder are bound, wherein the number of windows in the layout is consistent with that of the playing windows of the decoder. The method supports the simultaneous playing of a plurality of monitoring videos: the system supports simultaneous playing of different monitoring point videos on a television wall, each display unit can correspond to one decoding channel, and a single display unit can respectively display a single picture or 4 picture segmentation according to different configured decoders. And manual switching and custom polling switching of video of the television wall are supported. And the alarm window pane setting is supported, and a certain window pane can be designated as the alarm window pane and used for displaying the alarm linkage picture. Live wall-up is supported.

3.5 front integrated unit for detecting the invasion and the arrangement of people and the vehicle trip wire in the video and sending out alarm signal.

The video management cloud component supports the management integration of the intelligent camera and provides the following behavior analysis functions:

(1) and intrusion detection, namely automatically detecting intrusion behaviors in a set area in a monitoring area and giving an alarm to the intrusion behaviors. And the method supports multi-target simultaneous intrusion detection, can mark the intrusion direction of a target and gives an alarm for prompting.

(2) And (3) loitering detection, which is to automatically detect the staying and loitering behaviors of personnel in a set area in a monitoring area, alarm the staying and loitering behaviors, support the setting of a polygonal monitoring area and trigger the alarm only when a target in the area enters the monitoring area. The loitering and retention time setting is supported, the warning is not triggered when the target leaves within the set time, and the warning is given when the retention time exceeds a set threshold value.

(3) And the tripwire detection means that a warning line and a warning direction are set in a video monitoring view field, and the tripwire behavior in video monitoring is detected and an alarm is generated.

In some optional embodiments, the video analytics cloud component comprises:

and 3.6, the face recognition unit is used for extracting the features of the face data in the video or the picture to form a face feature database so as to search in the face database through the face picture.

The face recognition means that a face feature database is formed by extracting features of face data in a video or a picture so as to be convenient for retrieval in the face database through a face picture, and people who have passed through a real-time analysis video can be checked through face control so as to give an alarm for a hit face.

Face recognition supports a face recognition picture mode and a face recognition video mode.

The face recognition supports the real-time analysis of the face to obtain a face picture and face features, and the face picture and the face features are stored in a passerby library in real time; the method supports face analysis of historical videos, extracts face pictures and face features and enters a passerby library in real time.

Wherein the face recognition unit includes:

and the face real-time control unit is used for carrying out face real-time snapshot through the front-end camera, identifying the snapshot face and storing face information into the face feature resource library. The face real-time deployment and control unit supports face real-time snapshot through a front-end camera, identifies the snapshot face and stores face information into a face feature resource library. The system supports the establishment of a blacklist, and the facial features of the photos of key personnel are extracted and imported into a computer database. The number of the blacklist library is 10 ten thousand. Analyzing the face in the video in real time, comparing the detected face with the blacklist, and generating an alarm by comparing the hit face; supporting to increase, delete, modify and check blacklists and managing in groups; and the retrieval of the historical alarm is supported.

And the face dynamic library retrieval unit is used for traversing and comparing the uploaded face data with the face characteristic data in the face snapshot library and displaying a comparison result. The face recognition system automatically extracts face feature data in a video or a face picture extracted by a client, the face feature data and the face feature data in the face snapshot library are subjected to traversal comparison, and finally a comparison result is displayed by the platform. The face retrieval supports face snapshot in a real-time video, and face pictures and face features are stored in a passerby library in real time; the method supports face retrieval on the passerby library, and the retrieval result supports sequencing according to the similarity and time.

The face static library retrieval unit comprises a witness comparison unit and a static library retrieval unit, wherein the witness comparison unit is used for calculating the similarity according to two input face pictures, and the static library retrieval unit is used for returning the face pictures with the similarity above a threshold value according to one input face picture. The face static library is a personnel information library mainly used for recording less changes, such as a permanent personnel library, a household registration personnel library and the like, and is convenient for criminal investigation personnel to search detailed personnel information according to face pictures. The system supports the addition, deletion, modification and query of face pictures and personnel information; the system supports the batch import of face pictures and personnel information into a face library, and the face pictures and the personnel information are quickly uploaded and imported into a face system for face retrieval and personnel control functions.

Face 1:1 comparison (also known as witness comparison)

1:1 comparison of human faces, wherein the system can support a user to input two human face pictures, and the system returns the similarity for judging whether the two human face pictures are input by the same person system or not so as to carry out human face comparison; the system supports inputting a face picture and comparing the face picture with a certain certificate picture in a certificate library (static library).

Face 1: n comparison (static library search)

The system supports inputting 1 face picture, searching the face pictures with similarity higher than a specified threshold value in a specified static library, returning the search result by the system, and sorting according to the similarity. The system supports 64 static libraries at maximum, and the total number of the static library data supports 1 hundred million pieces at maximum.

Human face red/black/white list library:

the system supports the addition, deletion, modification and query of red/black/white/list face pictures and personnel information; the system supports the batch import of red/black/white list face pictures and personnel information into a face library, and the face pictures and the personnel information are quickly uploaded and imported into a face system for face retrieval and personnel control functions.

The system supports various list control and flexibly controls a control strategy; the person red list refers to persons needing privacy protection, VIP protection persons and the like, the red list is used for retrieval and filtration, and for the persons in the red list, a retrieval result cannot be returned to protect the whereabouts privacy of the persons in the red list; the red list is used for deployment and control filtering, and for the personnel in the red list, even if the personnel are in the deployment and control list, the deployment and control alarm cannot be returned. In different scenarios, the filtering function may be configurable, either just search filtering or control filtering or both may be selected. Application scenarios such as, for example, search or deployment for high security level personnel are not possible.

And the white list of the personnel refers to legal personnel, registered personnel and the like, is used for confirming the validity of the identity of the personnel, is the legal personnel of the identity in the white list, pushes the matching result of the face-passing picture, can configure the alarm rule, and pushes the successfully matched or unsuccessfully matched personnel to a third-party system as required or all the personnel are pushed.

And (4) personnel blacklist-control personnel list is assigned, personnel in the blacklist compare the face pictures, and alarm is generated if matching is performed, so that other subsequent systems can process conveniently.

And 3.7, the human body image searching unit is used for searching the video information through the human body image so as to locate the target human body information.

Searching the image by the human body means that a user can search the video information through the human body image, so that most of information in the video information is filtered, and the target human body information is quickly and accurately positioned from massive videos. The human body can quickly carry out global search in the cross-camera according to the color distribution and the posture characteristics of the target garment by searching the images, find out similar targets, then output the results in a snapshot form and sort according to the similarity. The human body searches for the picture with the picture suggestion scene has the big safe city of personnel management and control demand or important garden, the key feature:

the common camera is set to be used for controlling the cameras in a picture searching mode, so that the flexibility of the system is improved, and the cost is saved; the personnel information is searched fast and efficiently, and efficient management and control on the people are achieved.

The main functions of searching the picture by the human body are as follows: the human body is supported to search images according to images, a single image is supported to be input, similar targets in a target library are searched according to image characteristics, and the similar targets are sorted according to the similarity. The method supports the extraction and the indexing of the characteristics of searching the images of the real-time video/historical video and the uploaded video; fragment feature extraction and indexing are supported for the historical video and the uploaded video; extracting personnel characteristics (the supported human body attribute fields comprise gender, age group, jacket style, lower clothes style, upper body texture, hand-held objects, backpack, upper body color and lower body color); the method supports retrieval of a picture library which is captured by a human body checkpoint, and input pictures can be captured pictures of the human body checkpoint, uploaded pictures and video screenshots; the method supports human body feature extraction of human body bayonet deployment, supports non-motor vehicle and human body feature extraction of virtual bayonets, and supports vehicle and human body feature extraction of a human-vehicle mixed scene.

The human body is defined by the image searching characteristics and comprises the following steps:

sex: male and female;

age group: young, middle-aged, and elderly;

color of the upper part of the body: 11 dominant colors (black, green, blue, white/grey, yellow/orange/brown, red/pink/purple);

lower body color: 11 dominant colors (black, green, blue, white/grey, yellow/orange/brown, red/pink/purple);

upper body texture: pure color, horizontal bars, vertical bars and lattices;

the style of the jacket is as follows: long-sleeved and short-sleeved clothes;

the following clothes style: trousers, shorts, skirts;

carrying an object by hand: presence or absence;

backpack: presence or absence;

the riding state is as follows: yes, no.

And 3.8, a behavior analysis unit for performing perimeter intrusion detection in the monitoring area and giving an alarm to the intrusion behavior when the intrusion row is detected in the set area.

The video analysis cloud component supports a behavior analysis function, replaces the customer manpower to automatically detect key information in the video, reduces the customer manpower cost, improves the efficiency, and different intelligent analysis is suitable for different application scenes. Recommending scenes: there are important campus scenarios of intelligent monitoring needs.

The perimeter intrusion detection is to automatically detect the intrusion behavior in a set area and generate an alarm for the intrusion behavior in a monitoring area. The monitoring system is used for monitoring whether a target intrudes into the monitoring area. The monitoring personnel observe the video picture and send personnel to process, thereby avoiding loss. And the method supports multi-target simultaneous perimeter detection, can mark the target intrusion direction and gives an alarm for prompting.

The user can use the perimeter intrusion function to carry out the analysis to the video recording, whether have personnel to invade the appointed area in the quick monitoring video recording, improves the efficiency of looking up the analysis video recording.

The HASEN container cloud technology is a lightweight distributed resource management scheduling system. The method has the following characteristics:

light-weight clouding: the small-scale cluster is efficiently managed, the resource occupation on the computing node is less than 100MB, the distributed selection master and the distributed cache are embedded, a third party is not required to be relied on, and the performance is stronger; the light communication bus optimizes the communication performance; shielding the difference between different hardware and the bottom operating environment;

the deployment is convenient: compared with a virtualization technology, the container technology improves the utilization efficiency of memory, storage and bandwidth resources; in the running process, because the volume is small, the container can usually be started in the second level, the virtual machine needs to call a large amount of resources, and the starting time is in the minute level.

The management is convenient: the container technology can complete the management of a single container through a simple command line, and complete the quick packing and migration of the mirror image; meanwhile, the management of large-scale container clusters can be realized through the arrangement tool.

Elastic expansion: the container units are mutually independent and managed by a uniform arrangement tool, and the arrangement tool has the function of finding container nodes, so that the elastic expansion of the container can be automatically completed in a short time; meanwhile, each container is an independent individual, and the use of the resources called by the container and the use of the container are managed by the arranging tool, so that the use of the whole container system is not influenced by a certain container node;

high availability: similar to elastic expansion, when a certain container node fails, the container arrangement tool can find the change of the node in time and make adjustment in time according to the external request condition, the use of the whole container system is not influenced, and the high availability of the system is realized.

The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present application should be covered within the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims (8)

1. A traffic tunnel analysis platform, comprising:

the basic hardware layer adopts a cluster server architecture;

the system comprises a basic cloud platform, a HASEN container, a cloud service platform and a communication platform, wherein the HASEN container is adopted to provide a lightweight distributed resource management and scheduling system, and a management system based on clouding, servitization, opening unification and telecommunication level is constructed;

the video cloud component comprises a video management cloud component and a video analysis cloud component, the video management cloud component is used for managing a monitoring process and providing processing for real-time monitoring data, and the video analysis cloud component is used for identifying the acquired images and analyzing personnel and personnel behaviors.

2. The traffic tunnel analysis platform of claim 1, wherein the cluster server architecture comprises 4U two-way or one-way storage rack servers and 2U2 rack servers.

3. The traffic tunnel analysis platform of claim 1, wherein the application lifecycle management module comprises:

the IPC adding unit is used for loading the newly added IPC to the cloud node with lower load according to the load of the current cluster member;

the cloud node regulating and controlling unit is used for newly increasing or reducing cloud node members;

and the IPC adjusting unit is used for sharing the IPC load on the fault cloud node to other cloud nodes and returning the migrated IPC load to the original cloud node after the fault cloud node is repaired.

4. The traffic tunnel analysis platform of claim 1, wherein the video management cloud component comprises:

the video system management unit is used for responding to the user request and setting and changing the communication parameters, the monitoring parameters and the authority parameters of the video equipment;

the real-time monitoring unit is used for responding to a user request and setting and changing the style parameters of the video monitoring terminal;

the video management unit is used for responding to a configuration request of a user for video recording of the camera, recording the video and processing video data;

the television wall control unit is used for controlling decoding display of images on the plurality of large-screen liquid crystal display units;

and the front end integrated unit is used for detecting the invasion and the arrangement of personnel and vehicle tripwires in the video and sending out an alarm signal.

5. The traffic tunnel analysis platform of claim 4, wherein the video system management unit comprises:

the equipment management unit is used for providing access and disconnection to the front-end equipment;

the communication management unit is used for connecting external domain equipment and providing an external domain information configuration port for a user with management authority;

the front-end equipment monitoring parameter adjusting unit is used for setting and changing the parameters of a preset position, a guard position and a cruise track of the front-end video equipment;

and the user parameter management unit is used for managing roles, users and authorities.

6. The traffic tunnel analysis platform of claim 4, wherein in the real-time monitoring unit, the style parameters of the video monitoring end comprise window layout, split screen and display scale.

7. The traffic tunnel analysis platform of claim 1, wherein the video analytics cloud component comprises:

the face recognition unit is used for extracting the features of face data in the video or the picture to form a face feature database so as to retrieve the face data in the face database through the face picture;

the human body image searching unit is used for searching the video information through the human body image so as to locate the target human body information;

and the behavior analysis unit is used for carrying out perimeter intrusion detection in the monitoring area and giving an alarm to the intrusion behavior when the intrusion row is detected in the set area.

8. The traffic tunnel analysis platform of claim 7, wherein the face recognition unit comprises:

the human face real-time control unit is used for capturing a human face in real time through the front-end camera, identifying the captured human face and storing human face information into the human face feature resource library;

the face dynamic library retrieval unit is used for traversing and comparing the uploaded face data with face feature data in a face snapshot library and displaying a comparison result;

the face static library retrieval unit comprises a witness comparison unit and a static library retrieval unit, wherein the witness comparison unit is used for calculating the similarity according to two input face pictures, and the static library retrieval unit is used for returning the face pictures with the similarity above a threshold value according to one input face picture.

Images