Research On Technologies Of Intelligent Surveillance System Of Abandoned Object Detection

With the rapid development of video monitoring system, video surveillance in dynamic scenes, especially for humans and vehicles, has become one of the most active research topics in computer vision. It has a wide spectrum of promising applications, including access control in special areas, human identification at a distance, crowd flux statistics and congestion analysis, detection of anomalous behaviors, and abandoned object detection, etc.This thesis focuses on the key technologies of object detection and tracking and gives more attention to the enhancement of real-time and robust performance. The ultimate goal is to develop a practical intelligent surveillance system of abandoned object detection. The main contribution consists of three parts as follows: First, after a brief review of background subtraction based on the Gaussian mixture model, the update rate have been improved which can building background model fast at the initial moment with the bigger rate and can prevent the foreground from becoming to background with a smaller speed. The issues of moving shadow and ghost shadow in motion detection are also studied here and given solutions respectively. Second, in the proposed approach, occlusions are detected by analyzing MBB overlapping feature in consecutive frames, objects are tracked through combining mean shift and particle filter algorithm. Particle filter algorithm can handle occlusions between objects and mean shift algorithm is fast. So the algorithm after combined is robust to noise and improves the performance in real time. Last, Template Matching work is very important. It can be used to find a picture in the presence with the known template image and get the position.A novel approach is proposed which aim to detect abandoned objects. This approach is robust to noise and improves the performance in real time.