Research Of Moving Object's Detection And Tracking Based On OpenCV

Moving object detection and tracking is a hot problems in the computer vision, it has huge application demand in the military, medical diagnosis, unmanned, human-computer interaction and other fields. Analyze the advantages and disadvantages of the algorithm through the study of target detection and tracking algorithm which is used at home and abroad commonly,aiming at the defect of the traditional Mean Shift tracking algorithm that need to select target window artificially, put forward an moving target automatic extraction and tracking algorithm. This algorithm using gaussian mixture background modeling algorithm in the target extraction phase,and improve the using mechanism of the two parameters of learning rate alpha and threshold T of background model algorithm, use matching feedback quantity to adjust the parameter adaptively. In target tracking phase,combine Mean Shift algorithm with improved gaussian mixture background modeling algorithm, fistly,process the foreground image which produced by improved gaussian mixture with morphology processing method,then segment the target area, finally identify the tracking window through calculating the area of the candidate target in division area. Using OpenCV open-source computer vision library and VS2010 to form experiment platform, the experimental results show that the improved gaussian mixture model make up for the deficiency of the traditional gaussian mixture model, enhances the model's ability to adapt to complex monitoring scenario, improved Mean Shift algorithm can automatically select the tracking object, make up for the traditional Mean Shift algorithm need select target artificially, can realize automatic tracking.