Key Technology Research And Implementation Of Intelligent Video Analysis System

Video of intelligent analysis is research hot in the field of machine vision . It is including image processing, target tracking, target identification, image reconstruction, etc.In this paper, research on the subject is key technology research and implementation of intelligent video analysis system that is in great demand. To design of different algorithms for different scenarios and different scenarios in research, in order to achieve reference and use for the relevant departments ,and provide valuable reference for the same field of study .During the course of the study subjects, involving the main contents of the following:First, this article studied the location of the intelligence analysis system at the monitoring system, using the system architecture map to embody its advantages of location. And according to the workflow of the intelligence analysis system, study the basic concept and basic algorithm. It can provide basic theoretical foundation for algorithm design of specific scene.And then, According to the workflow of the intelligence analysis system, in-depth study the key technologies of background modeling and extraction of the goal and to identify suspicious goals. First of all, against the background of the modeling techniques to analyze, to identify and apply to their respective advantages and disadvantages of different scenes, put forward to the same pixel of the background modeling algorithm, and detailed study the background initialization algorithm of approximate color mode ?s average and the background update algorithm of local color combination of time and noise. Second, focus on the object extraction algorithms, according to the three parts: binarization processing of differential background and static image of the pre– and goal of separation, study and put forward to the adaptive threshold and the denoising algorithm of closing and opening and the goal of separation algorithm of four neighborhood. And study the technology how to choose a goal to distinguish the characteristics. Final, according to the ideas of the scene to determine the suspicious target, study the algorithms of campaign to change state and object tracking to identify suspicious goals, one proposed method of to identify illegal parking at highway, one proposed method of goal match. Specific study the design of kalman filter input parameters and algorithm of follow-on target of matching function.In the process of to study algorithm, this paper give the achieve result of algorithm designed. And according to the result, analyze the problems, put forward to the solutions and the focus for follow-up study. Provide the reference information for engage in research.