Detection and Tracking of Human Motion Targets in Video Images Based on Camshift Algorithms

For the traditional moving target tracking algorithm, it is difficult to detect the target area accurately because of the influence of environmental factors. Therefore, manual selection of moving targets is often used, but this method is not intelligent enough, and has strong subjectivity and experience, so the data obtained is not convincing enough. For this reason, based on the multi-frame subtraction method, the connected region search is used to select the appropriate moving target and calibrate the target area, and then the moving target is determined and tracked. At the target tracking stage, the fusion feature vectors in the calibrated area are updated in real time to improve the accuracy and robustness of the follow-up tracking algorithm. Experiments show that the method is feasible. In the stage of target detection, the improved algorithm filters and calibrates the target area on the premise of increasing a small amount of computation. In the stage of target tracking, the computational area is reduced to the calibration and tracking range, which improves the efficiency of the algorithm and guarantees the stability and robustness of the algorithm.