Industry News
    Home > News
  • Video analysis technology application three types of status quo

        Intelligent video analysis technology in the field of pattern recognition, artificial intelligence , intelligent algorithm embedded into the DSP , through analysis and extraction of personnel and vehicles of various patterns of behavior II objectives form the core algorithm. For now, though in recent years, the rapid development of intelligent video technology , and the limitations of intelligent video analysis algorithm itself , in an ideal environment in order to achieve its full functionality , so in practice , only the initial realization of simple functions. But the strong market demand , so why intelligent video analysis technology can not flex its muscles ? Future intelligent video analysis technology which in turn will have a breakthrough ?
        Applications are currently three types of video analysis technology status quo
        From a technical perspective, the current domestic intelligence analysis is also concentrated in two main categories: one is the use of split screen foreground extraction methods such as video footage of target extraction testing, through a variety of different rules to distinguish different events in order to achieve a different judge and a corresponding alarm linkage , for example: some of the earliest functional behavior analysis ( cross-border , regional invasion, fights detection, people gather , etc.) , as well as early detection belong to this type of traffic incident algorithm technology. Another is the use of pattern recognition techniques to screen a particular object modeling , and training through a large number of samples , so as to achieve the video screen to detect specific objects and related applications . Such as vehicle detection , face detection , head detection ( flow statistics ) applications.
        From the application point of view, the current domestic intelligence analysis there are four main categories: The first category is a real-time alarm . The second category is the statistics class , third class is the class attribute recognition . The fourth category is the image processing category.
        The form of products , at present there are two main categories of products on the market form , one is intelligent front-end products . One is the back-end server products. These two products have advantages and disadvantages , depending on the application and project types have different options.
        Currently on the market you can see a lot of smart camera , some video analysis algorithms will soon migrate to IP camera , the realization of real-time video analysis and inspection in the camera , to achieve some of the intelligence analysis function ( for example, currently has a vehicle to achieve vehicle detection cameras capture achieve regional functions such as intrusion detection alarm camera target , there is a camera to capture the realization of human detection , etc. ) , the camera comes with intelligent analysis , system architecture is simple, while the current camera DSP processing capacity has been fully able to run a variety of more complex analysis algorithms , analysis and back-end product is basically the same effect .
        Intelligent Video Analysis is the most mature technology called Intelligent Monitoring System
        Monitoring platform development these years , has been very clear definition close monitoring , integration trend , but the application of intelligent performance has just begun , but the outlook is not small. We say surveillance intelligence, many industry insiders believe that intelligence is monitoring intelligent video analysis . Judging from the current situation analysis applications , intelligent video analysis technology is the most mature technology of intelligent monitoring technology system .
        Currently intelligent analysis technology has been widely used in various security fields , distinguished according to the way of its implementation , the industry analysis to summarize it as the following types of intelligence.
        Intelligent analysis diagnostic categories : diagnostic categories appear intelligent analysis mainly snow for video images, scrolling , blur, color cast, gain imbalance , PTZ control , screen freezes and other common camera malfunction , video signal interference, decreased video quality for accurate analysis judgment and alarms. Diagnostic class intelligent analysis technology is relatively simple to implement , usually in the form of client management platform after the emergence of large-scale monitoring of projects, in particular the role of the daily operation and maintenance of city-level monitoring is very obvious. Some currently available DSP- based intelligent analysis equipment , DVR and DVS and so comes the auxiliary functions.
        Recognition of Intelligent Analysis: The technology bias in the analysis of the static scene processing , image recognition , image matching and pattern matching , and other core technology, extraction and analysis of people, vehicles , material and other relevant characteristics of information. In recognition analysis on the car license plate recognition technology is mainly applied . After years of development of the technology and applications , has been very mature. Blurred license plate recognition technology enables the reduction and the technology is no longer confined to high-definition , standard definition field began to spread . License plate recognition technology has been widely used in the car park entrance , highway toll stations , etc. , is developing rapidly in recent years : electronic mount system with traffic , license plate recognition technology has been used extensively for vehicle traffic violation snap , effectively reducing the the number of vehicular traffic violation , greatly reducing traffic accidents.
        Behavior class intelligent analysis : The technology focuses on the dynamic scene analysis. Typical features are: retrograde vehicles and related traffic violation detection, intrusion detection zones , fences crossing detection, tripwire crossing detection, articles theft detection, lane detection and traffic statistics and other operations . Motion detection (VMD) is a kind of intelligent analysis of the " early intelligence ", VMD video screen changes according to the movement of pixels in a block to determine shortcomings obvious : Based on the two-dimensional analysis, false positives too , does not recognize the move pixel block is interference or goal.
        For now monitoring the application of the most sophisticated intelligence intelligent video analysis technology point of view, it can in the case of almost no human intervention , through the camera recording the image sequences for automatic analysis of dynamic scene to locate the target identification and tracking , and analyzing and judging the behavior of the target based on this , so both can be completed at the time of the daily management of abnormal happens to react promptly , so as to solve the traditional monitoring workload, low efficiency, reaction slow and other issues , the advantage is obvious , but also to promote one of the most critical technical entire security intelligence .