Research Of Some Key Algorithms In Intelligent Digital Video Surveillance System

Nowadays, video surveillance system is playing an increasingly important roll in all ways of our life because of its visibility, convenience and the abundance of content. In order to solve those inadequacies including automatic detection of moving object, the object occlusion by other objects or disturbance by multiple close colors and real time in most existed monitoring systems, this thesis presents a real-time digital video surveillance solution with intelligent detection in complicated dynamic background.The main contents of this paper are shown as following:1. Implement the real-time video image capture and introduce a method of fast decomposing AVI data stream by using the functions in the VFW library. Now we have solved the input of entire video surveillance system. Based on object-oriented program thinking, introduce general method of serial communication and implement the console controlling used PELCO-P protocol.2. Implement moving object detection based on Canny edge detector after analyzing some kinds of detection algorithm. Solve the remnant background in moving object edge image using judgement of Sparse and Dense advanced. Extract the boundary of moving object for the following tracking based on mathematical morphology.3. Based on the mode of improved histogram back-projection and searching method of modified Mean Shift to deal with dynamically changing color probability distributions derived from video frame sequences, implement fast object tracking.4. To deal with the object occlusion by other objects or disturbance by multiple close colors, a two-dimensional Kalman filter is used to predict the object region centroid and modify the value of estimation based on actual movement parameters, which is proved robust and efficient.Very promising experimental results have confirmed that the proposed intelligent digital video surveillance system is efficient and reliable for moving object detection and tracking.