Research And Improvement Of Intelligent Video Surveillance In Moving Target Detection And Tracking

Intelligent video analysis is an important research direction of computer vision, Has the important application value. Moving object detection and tracking is the basis for intelligent video analysis, The main content of this paper is the moving object detection and tracking.Based on the algorithm of moving target detection is mixed Gauss background modeling, is currently the research and application of the moving target detection algorithm is widely; the main idea is:the assumption that each pixel will follow a Gauss distribution, but due to noise and moving target, a single probability distribution is not enough to simulate the Gauss single pixel in time series the distribution of mixed methods emerge as the times require, so Gauss. It is found in the experiment, using multiple hybrid Gauss distribution is also difficult to accurately model is established for a single pixel, for example rapid natural light changes, the small targets and shallow shadow as well as random oscillation problems; these are the problems to be solved in this thesisThrough literature review and analysis of the experimental data, make full use of the pixel correlation in time and space, the above problems in the moving target detection is improved, the robustness and accuracy of the algorithm are improved, which can provide more accurate for the moving target tracking system.Moving target tracking algorithm based on the Mean Shift target tracking algorithm, Mean Shift algorithm based on the experimental findings, not effective in fast tracking; through theoretical analysis and data comparison, search the Mean Shift algorithm in target, the search window is limited, which can’t effectively tracking fast moving targets. To predict the target position, Calman filtering can therefore, in the Mean Shift tracking algorithm, the introduction of Calman filter forecast, tracking in the forecast, thus effectively keep up with the situation. The tracking effect is improved.This paper firstly introduces the application of intelligent surveillance system scenarios and research significance; Secondly, a comprehensive description of the moving target detection and tracking algorithm of moving target detection algorithm of moving object modeling problems; third chapter of mixed Gaussbackground, moving target tracking based on mean shift algorithm; in the fourth chapter, the moving target detection and tracking problems are put forward some improvement methods; the fifth chapterthe main experimental contrast effect, in order to see the visual effect of the improved; finally is the summary and outlook.