Foshvision Technology Co.,ltd

Foshvision can supply all high quality security night systems and thermal systems!

Home > News > Content
The Development Of Intelligent Video Analysis Technology
- Feb 19, 2009 -

Along with the input HD probe, people's demand for intelligent video analysis technology more and more, have an increasingly high demand, the security industry to bring a wider range of ideas, many intelligent rapid product emerges:
1, stereo technology
Binocular stereo technology core purpose is to improve recognition accuracy. Due to the formation of stereo vision technology with the geometry of the object information in the field of view, so that the sets of rules in effect, excluding interference factors such as light, shadow, and significantly improve the accuracy of intelligent analysis. If the HD technology by improving the available pixels to increase the accuracy of analysis, tactical moves, the binocular stereo vision technology is a strategic impact on the accuracy of video analysis. Binocular stereo vision technology is based on the principle of Parallax, and imaging device from different locations to get the two images of the object to be measured by calculating the position error between the corresponding points of the image, to obtain the three dimensional geometric information of object methods. Dual camera or camera space within the Visual field of the free movement of three dimensional coordinates and attitude make high precision measurements, determine the mass of moving objects, and according to the results of calibration for high precision tracking moving targets.
Stereo vision technology to track, due to the ability to identify the coordinates of the target, posture, distance, space distance and background, so it can adapt to the complex tracking context. Binocular technology applied to the body property identification is a technical application of face recognition technology in step, the more accurate positioning and analysis of characteristics of people supporting an important role.
2, ball joint tracking technology
Linkage ball machine tracking technology uses intelligent track ball technology as the Foundation. On the application level, common single point of the track ball can be monitored, promoted to seamless relay tracking systems within a single target, used with map, can easily achieve seamless tracking for high security level area and achieve the target picture, crime alerts and other high level security requirements. Realization of linkage ball machine tracking technology, you need the multiple target recognition and tracking technology. In the application, usually sets a ball as a launching point, for intelligent behavior analysis of wide area within the target, and simultaneously monitor multiple targets sorted according to established policies, and in accordance with the order, directed by tracking and monitoring intelligent track ball machine target. Compared with single target tracking, multiple target tracking crucial point is the data Association problem, namely the establishment of a unified coordinate system allows the machine to coordinate information to the tracking of the target machine realization linkage tracking.
3, after the application of intelligent technology
With the popularity of surveillance cameras, surveillance systems there is a large amount of video data in the current manual viewing mode, the traditional method is required to order from start to finish, often takes several times that of the original video to shenkan completed, requiring large numbers of video shenkan working overtime for weeks in a row. In order to avoid omissions and errors, it is necessary to increase investment in human. But practice proves that this thankless job, still not solve the underlying problem, how effective and efficient applications, reduce the inefficient manual viewing playback, high cost, and fatigue, and in different resolutions, different resolution video to identify exactly what you want to get information, based on the above requirements, security manufacturers developed a summary of the video, video search technology.
Video abstraction techniques
Video summary form video pieces, different goals "through time and space" at the same time show playing, making the 24-hour video was made into a short summary to minutes video becomes a reality. Video summary not only concentrates the essence of the event, all events, no value of the video will be removed. By snapshot technology, you can watch all the activities possible in a few seconds, back to the original video, instant lock on its target position in the original video. These intelligent video analytics implementation and application will greatly improve the efficiency of mass analysis of the footage.
Video retrieval technology
Video retrieval is mainly dependent on video preprocessing algorithm for video, through the structured processing of video content, extracts valid information in the video content, marking or treatment, people can search through various properties description. Is the use of video detection algorithm for video retrieval the main structured description of the video, now in the algorithm used in the product are mainly the following: behavior analysis algorithms, license plate recognition algorithms, color recognition algorithm of vehicle, vehicle identification algorithms, vehicle identification, face recognition algorithms, human feature recognition algorithms, etc. The human feature recognition including the person's age, gender, height, clothes, colors, whether you wear glasses feature information such as identification. Video retrieval algorithm has been applied technology is behavior analysis algorithm, license plate recognition, vehicle color recognition algorithm, face detection and recognition algorithms. Due to the arrangement of cameras will have blind spots and other reasons, when stereo technology and ball tracking technology cannot accurately capture the moment the suspect while trajectory information, you can probe video for later retrieval and analysis of on-campus and find clues to help investigators quickly locating suspects in a timely manner.
4, video stitching
Video display system is based on image stitching technology achieved, image mosaic based on actual scientific research and engineering development on the needs of the. In many fields often use more than the human eye view high resolution images, perspective often fail to meet the needs of ordinary cameras and camcorders, for example due to distance limitations, some oversized objects cannot clear video filmed.
Currently for, many large application scene are need high clear high cover of shooting, like airport runway, and Terminal,, a camera cannot real clear of show these large application scene of panorama, cannot to out a user satisfaction of HD video, more cannot on video in the of things for HD analysis, which and brings of is high input and more picture show, not only visual effect bad, also cannot maximum degree of meet user of HD big picture of needs. Based on the above requirements, image mosaic technique to solve this problem, the technique will be images spliced together from different angles to get high resolution images, it solves the urgent demand for high-definition monitor user scenario, users can browse the high-definition picture in a video image.