Analysis of the characteristics of intelligent monitoring technology of public security bureaus
[Abstract] Most of the monitoring of the current supervisory offices in China is still dominated by traditional monitoring. Despite the increasing popularity of video surveillance systems, most of them are viewed through live monitoring. Since human attention is difficult to concentrate for a long time, the effect of manual monitoring is drastically attenuated as the event grows. Therefore, despite knowing some key scenarios, public security officers are unable to monitor them at any time. Although a small part of the use of intelligent monitoring, but the actual application of intelligent monitoring is not very good. In order to solve such problems, intelligent video application technology emerged as the times require, mainly including real-time monitoring and video retrieval. The real-time monitoring part mainly includes intelligent analysis technology and video diagnosis technology. Video search technology mainly includes key marking technology
Introduction
Most of the monitoring of the current supervisory offices in China is still dominated by traditional monitoring. Despite the increasing popularity of video surveillance systems, most of them are viewed through live monitoring. Since human attention is difficult to concentrate for a long time, the effect of manual monitoring is drastically attenuated as the event grows. Therefore, despite knowing some key scenarios, public security officers are unable to monitor them at any time. Although a small part of the use of intelligent monitoring, but the actual application of intelligent monitoring is not very good.
In order to solve such problems, intelligent video application technology emerged as the times require, mainly including real-time monitoring and video retrieval. The real-time monitoring part mainly includes intelligent analysis technology and video diagnosis technology. The video retrieval part mainly includes key marking technology and feature recognition technology.
I. Problems to be solved by the intelligent monitoring of the supervisory office
The problems faced by the current intelligent monitoring of the supervisory office are mainly due to false positives and false negatives, which makes it difficult for the police on duty to face the police. It objectively increases the pressure on the police. The intelligent monitoring of the supervisory office needs to solve the following main problems:
(1) Solve the problem that the police on duty use the naked eye to monitor and find the problem, and use the mode of providing the early warning to let the police on duty do not have to stare at the monitoring at all times, and only pay attention to the pop-up screen when there is an early warning.
(2) Set different warning and alarm mechanisms for different areas and different time in the supervision office.
(3) Increase the mechanism of analysis and judgment to improve the accuracy of early warning and alarm.
(4) Provide emergency plans for emergencies for timely disposal.
(5) Make full use of the network architecture to realize timely delivery of video images and early warning (alarm) information.
Second, the intelligent monitoring solution technical features
Its technical features include: intelligent analysis technology, video diagnostic technology, key marking technology, feature recognition technology, and intelligent search technology.
(1) Intelligent analysis technology
Intelligent analysis technology supports various intelligent applications and extensions, and can realize functions such as regional alert, crowd density, image anomaly, human traffic statistics, face monitoring and even face recognition comparison. By combining the standards of the supervisory business standards, the information is classified and processed. Provide intelligence information analysis services for prevention and control management. The main application features are as follows:
Regional alert: Automatic monitoring and alarming for personnel movement in the user-defined area, such as detecting whether a target enters a predetermined area (or crossing the line) and prompting an alarm. In this technology, crowded aggregation monitoring can be expanded.
Behavior Analysis: Automatic monitoring and alarm for target fights or strenuous exercise.
Item Remaining/Item Loss: Automatically monitors and alarms when suspicious people are squatting or staying in the designated area for a long time, and automatically monitors and alarms when the rest of the characters suddenly disappear.
Human flow statistics: For the automatic monitoring of one or more entrances or exits or single direction/two-way crowd flow, provide a more accurate statistical analysis of the flow of people and the direction of flow of people. In this technology, the population density can be graded, and the threshold value can be set in time to avoid the occurrence of malignant gathering accidents.
Face detection: Face detection and tracking of targets in the set area, screening out face image real-time control, and providing basis for later query.
Intelligent analysis technology can be applied in a variety of ways, not only in the front-end camera, but also in the back-end management platform software, which can be flexibly configured according to the actual project requirements and cost budget. The advantages and disadvantages of intelligent analysis technology mainly consider the accuracy rate, which puts forward specific requirements for the selection and deployment of on-site cameras. For example, the flow statistics can be applied to gates, passages, corridors, etc., and the angle between the camera and the ground needs to be considered. The angle of the flow direction, the lighting conditions and the weather conditions (the sunny days and the rain and snow will have different effects).
(2) Video diagnostic technology
Video diagnostic technology can comprehensively evaluate various image quality in the monitoring system, give the quality evaluation result of the video, and make common camera faults such as snowflake, scrolling, blur, color cast, picture freeze, gain imbalance, etc. appearing in the image. Accurately judge and issue alarm information. The system can also detect the presence or absence of video signals and the operation of the front-end PTZ camera, effectively prevent image quality problems caused by hardware and unnecessary losses, and timely detect the illegal behavior of the damage monitoring equipment. The main test contents are as follows:
Signal Missing Detection: Automatic detection of intermittent or persistent video loss caused by front-end pan/tilt, camera malfunction, damage, vandalism or video transmission failure.
Sharpness anomaly detection: Automatically detects image blurring in the main part of the field of view due to improper focus, lens damage, or foreign object shadowing. Automatically detects when the lens is aimed at an innocent object.
Signal brightness anomaly detection: The screen is too dark, too bright or black due to camera failure, gain control disorder, abnormal lighting conditions or artificial malicious occlusion.
Video noise detection: Automatically detect video images with cluttered "crossbars", "ripples", or bursts of noise, distortion, snow, jitter, or scrolling caused by clutter, thorns, and line interference.
Video color cast detection: Automatically detects the color cast phenomenon in the video caused by poor line contact, external interference or camera failure, etc., mainly including full-screen single color cast or multi-color mixed strip color cast. Video diagnostic technology also includes a variety of application methods, which can be integrated into the front-end camera, but is more suitable for deployment in the back-end management software system. In the networked system, the use is more flexible, and all the videos in the network are regularly inspected and discovered in time. Failure, speed up the response of maintenance, and ensure reliable operation of the system.
(3) Key marking technology
All the cameras in the office are constantly monitored and recorded 24 hours a day. In the face of the massive video and video data generated, how to find useful clues in the shortest time to provide an example for solving the crime. The video recording is played back one by one through human tactics – not only inefficient And because of human physiological limitations, it is easy to miss important scenes and clues.
To this end, the associated information at the time of the event can be associated with the video record as a sign of the later search, for example, when manually recording, capturing, and making a recording plan, inputting a text prompt as a key mark; when the license plate is recognized, the license plate number is The captured photo is associated with the video clip; the person's access card identity information is associated with the photo and video when entering and exiting the channel; the alert information is associated with the triggered video. In short, a certain piece of information of a key event is associated with the video material, and the video is segmented by the keyword, and the event log and the video snapshot are assisted, so that the specific moment of the video file to be queried can be quickly located, which is convenient for later retrieval.
(4) Feature recognition technology
When manual and other physical detection technologies are unable to correlate video, the intelligent analysis technology of video content based on target features provides us with a wider application potential. Typical applications are face matching technology and smart search technology.
Face comparison system:
Based on the biometric recognition technology of the human face, the human face can be automatically captured from the video, and the similarity comparison is performed with the target library in the background system, and the comparison result is pushed to the monitoring responsible person. Regardless of whether there is a comparison result, the captured face will be stored in the back-end system. When there is a case that needs to check the monitoring data in the future, the face recognition technology can be used to confirm the suspect's action track, providing efficient technical support for solving the case. The face recognition system usually includes three steps of collection, comparison and processing. The acquisition link obtains the high-quality image of the object to be compared; the comparison link converts the image into a feature code that can be understood by the computer, and compares it with the pre-stored face database to confirm the identity of the object to be compared; the processing link compares the result Tightly linked to the user's business application.
The application mode of the technology mainly includes: verification mode, given two face photos, automatically determining whether it is a person, and ensuring the identity of the person. Search mode, given a face photo, compare it with all faces in the face database, to find out whether there is this person, and the returned result is the information of the personnel or the list of candidates in descending order of similarity. Inquiries and verification of personnel status.
Through the face comparison system, it is possible to obtain the past whereabouts information of the custody control personnel in time, and analyze and judge the intelligence personnel of the criminal investigation, the investigation, the anti-drug and other police personnel, and formulate a follow-up operational plan; when a new case arises, the police can use the police The face image of the suspect is checked in the log library to determine the whereabouts of the past period of time, providing clues for the detection of the case.
(5) Intelligent search technology
The intelligent search technology is based on the target model and feature vector extraction method. Through a target photo or even a witness description, the index condition of the target object can be generated. Using the classification search or the sample search, the tens or hundreds of target features are adopted. Thousands of times the speed of the search and analysis of video data, find all suspicious targets, find the target (or have a sufficiently high degree of relevance) possible recording location, generate a snapshot list to provide user search and playback.
The target features mainly include elements such as color, height, and direction. It can be expressed in terms of input semantics, or it can be queried by providing samples, such as a person wearing a red jacket or white trousers, and the user can input the selected details of the clothing to search. Based on intelligent search technology, it can search in parallel in massive video data and conduct multi-objective relationship analysis.
Flexible application of intelligent search and intelligent analysis technology can quickly respond and eliminate crime in its infancy. For example, the police station found that there were dangerous goods in the le gacy package of an access channel. Through intelligent search technology, the area was invaded and carried with parcels, and the 24-hour archive video of several cameras around the scene was inquired. A valuable piece of event, after manually confirming the suspect carrying the package, so as to locate the specific time, the system searches for the 48-hour video of the nearby main entrance and exit according to the image of the suspect’s scene, and gets it at a certain door. The more detailed feature image, and then continue to track the target trajectory from dozens of cameras in the surrounding area, thereby obtaining its route of travel and contacts, and finally the suspects are all captured.
Third, the typical application of the intelligent monitoring of the institution
The intelligent monitoring of the supervisory office integrates the digital surveillance platform of intelligent video processing technology, and realizes the application of image information by collecting various image resources. It can be integrated with the police information comprehensive application platform to achieve integration of information collection and comprehensive application. Through data docking and information fusion, image resources and data resources can be re-processed, and information retrieval and analysis can be used to realize high-risk early warning, personnel screening, case serialization, dynamic management and control, early warning control, and classified disposal. Provide comprehensive application and data support.
As a special place, the scope of supervision needs to be different from the general social monitoring. It is not applicable to bring some intelligent technical means for ordinary social monitoring directly to the supervision office. Practitioners who are truly committed to the security of the institution should develop intelligent monitoring technologies for the regulatory sites in conjunction with the specific environment of the institution.
There are many subsystems involved in the security system of the supervision site. It is necessary to consider fully sharing and linking information with each subsystem to make it work more effectively. At the same time, more consideration should be given to the integration of IoT technology and intelligent monitoring. Intelligent monitoring is ultimately applied to the management of the supervisory office. Only when it closely matches the daily management mode and working mode, can it maximize its role and ensure the safety of the institution.
English
Español
Русский
Deutsch
Français
Português
