Moving Target Tracking Method Research In Video Sequences

Moving target detection and tracking technology, as one of the important technologies in computer vision research field, includes advanced technology in computer, artificial intelligence, pattern classification, image processing and automatic control, etc. Moving target detection and tracking technology has wide applications in video monitoring, image compression, military target tracking and robot vision navigation, etc. In a complex actual scene, it is very difficult to ensure the reliability and real time of moving target detection and tracking.This paper studies moving object detection technology under the condition of moving camera. The robust multi-resolution algorithm used in this paper first builds model of camera movement, and then estimate the motion of camera through two continuous frames in video. If the motion of the pixel is the same with camera movement, the pixel is considered as a background pixel. If the motion of the pixel is inconsistent with camera movement, the pixel is considered as a foreground pixel which is a part of moving objects. The method in this paper can well solve moving object detection under the condition of moving camera.This paper studies robust tracking problems in various complex scenes and puts forward multi-cue-based CamShift guided particle filter tracking algorithm. The algorithm adopts both color and motion features and uses CamShift to optimize particle filter algorithm. Meanwhile, an occlusion strategy is used to deal with full occlusion for a long time. The proposed algorithm in this paper can realize reliable target tracking in many complex cases, for example: structural deformation of target; background clutters; illumination variation; occlusion for a long time, etc.