Research On Information Processing Algarithms And The Design Of Intelligent Video Surveillance System

Intelligent video surveillance systems, which involve many subjects such as image analysis, pattern recognition, artificial intelligence, and behavioristics, are widely used in numerous fields: public security monitoring, medical treatment and attention, traffic management, human-computer interaction, real-time analysis, et al. There are important theoretical significance and extensive application prospects for the research on information processing algarithms and the design of intelligent video surveillance systems. Therefore, intelligent video surveillance systems have become research hotspot in the fields of pattern recognition and artificial intelligence for domestic and overseas recent years.This dissertation takes intelligent video surveillance as research object and deeply analyzes the algorithms of video image processing and target detection, matching, tracking. It technically explores the approaches and implementation models of moving target detection and tracking, verifying the robustness, real-time and system feasibility of video information processing from the angle of the experiment and application. The research work mainly covers the following innovative topics:1. In the research of fractional differential image enhancement and image denoising, an approximate sixteen directions 5×5 fractional differential template is constructed according to vector synthesis theorem. The template is used to image filtering and the experimental results show that the algorithm has better effect to the enhancement of image texture.2. An algorithm of adaptive background modeling and shadow elimination based on color space conversion was proposed. The adaptive background difference was used to separate background and foreground of image and the detected target was locked in active contour(means rectangular box) for continuous target tracking in RGB space. The shadow detected along with the target was removed by using its brightness and other information in HSV color space.3. An intelligent video surveillance system of laboratory based on SOPC is designed and implemented. The platform of this system is Altera Corporation’s TE3 development board, equipped with the CCD high-speed camera and sensors, to achieve real-time image acquisition, processing, compression, and moving target detection and tracking.4. A dual cameras collaborative tracking system is designed. The main idea is to build a lookup table for dual cameras calibration, applying the algorithm of mixed gauss background modeling to detect moving target and applying Camshift algorithm to track the targets. The system can track moving target effectively and zoom the target, achieveing the goal of collaborative control of dual cameras and sustainable tracking.The dissertation combines with computer vision, signal processing, embedded system, computer network technologies, intelligent surveillance and other related theories, to do some beneficial exploration for image processing of intelligent video surveillance, target detection and tracking, and the design and implementation of intelligent video surveillance system. The purpose of the research work is to achieve a certain achievements of theory and practice, and to provide some reference and guidance for the design and development of intelligent video surveillance system.