Research On Moving Object Detection In Intelligent Video Surveillance System

Video surveillance system is an important part of public security. It is widely applied in various kinds of fields such as national defense, social security, traffic, electric power system and industrial control system. The detection technology of moving objects contains image processing, pattern recognition, artificial intelligence and other advanced technology of computer science. And it is becoming an important field of computer vision. Intelligent video surveillance technology enables computer having the human visual capacity to understand the meaning of dynamic scenes by using computer vision technology. Moving objects detection is the foundation and precondition of intelligent video surveillance technology.In a variety of moving object detection methods, background subtraction is a very important method which is widely used in moving object detection in static background. Because actual surveillance video is very complicated, the key problem of background subtraction algorithm is making the background model be more intelligent and more robust.An improved moving object detection algorithm based on background subtraction is proposed in this thesis. This thesis presents the conceptual design of making a combination of Gaussian mixture background modeling and running average background updating in order to decrease the computation cost of background modeling and improve real-time ability. Meanwhile, the background model can be more robust. Different updating factors are used for means and variances of each model to improve the detection sensitivity and to reduce the fitness of the model. An improved maximum between-cluster variance method is proposed by this thesis to enhance the adaptive ability in the moving object segmentation. Furthermore, after a series of post-processing methods, the moving objects areas are more integrated, and the profiles become clearer and more accurate. Then more accurate information of moving objects can be provided for tracking and intelligent analysis.