Research On Active Video Surveillance

Active surveillance is one of the most significant and high-tech applications in domain of computer intelligence video surveillance. it infers a lot of subjects including image engineering, pattern analysis, artificial intelligence. It can automatically analyze the sequence of images by the methods of computer vision and video surveillance. This system can defect, position, and track objects in a moving environment in real-time. Furthermore, it can also analyze and judge the movement of objects. The aims of this system are to understand the meanings of video stream and to explain the scenes comprehension to guide and make decisions.Although there are still many problems in computer intelligent video surveillance, it has been used in many areas. A lot of breakout was made in this area by the work of researchers of the nations. Based on existing conclusion; this paper describes research in the key technology of video surveillance.This paper mainly focuses on the research on moving object detecting and tracking in still background. At first, paper describes three algorithms in moving object detection: optical flow method, the instantaneous difference method and background modeling method and analyzes there advantages and disadvantages. Secondly, it indicates usual image preprocessing technology: image noise reduction ,image enhance, image restore and research the image graying methods such as average method, weighted average method, largest value method and energy method. Then, it gives a algorithm which is combined by frame difference method and area growing method. Firstly, difference method is used on adjacent frames. Then bilingual picture is reached by segmentation according to the given threshold, meanwhile we scan bilingual picture to get rid of lonely polluted pixels and false objects. Lastly, region growing method is used to improve the edge of the objects. Experiment indicates that the algorithm in this paper is more efficient in detecting objects in complicated background compared to consecutive frames difference method and mathematic morphology method. The algorithm makes the moving object tracking more precisely because it does not keep any background pixels. At last, contempt object tracking algorithms are categorized in the paper.