Fast Moving Object Detecting And Tracking Technology Based On Static Scene

Video tracking is an important research direction in the computer vision fields, it mainly refers to the moving objects detecting, classification, identification and tracking. Since the target tracking process is often effected by the target deformation, object occlusion, object removing, background environment change, ect. So the question of making the target tracking algorithm more robustness, more accuracy, less computer time and transfer more easily have become a challenging research topic.This paper choose target detecting and tracking contents in the static scene as the main research objects. The common target detection algorithm and tracking algorithm were studied and classified in the article. A object detecting algorithm based on the continuity of video detection algorithm was proposed, which had two innovation points:for the point of background reconstruction, a fast background reconstruction method was proposed. It's based on pixel intensity classification method and the assumption of background pixels can be observed in at least 50% time in the training phases. Compared with the traditional pixel intensity classification algorithm, this method effectively reduced the modeling time and at the same time ensured the precision; For the point of moving object shadow removing, an advanced method based on the difference of the proportion of the RGB was presented, It could remove the shadow of moving object quickly and effectively.For the target tracking problem, the color information of the moving object were selected as the object character. Kalman filter and Meanshift algorithm were combined together. It could achieve good tracking performance. The OpenCV and it's vision tracking framework were introduced. Through the embedding of the designed moving object detecting module, the fast and effective tracking was realized. Simulation and experimental results demonstrated the validity of the proposed method.