Pedestrian Target Tracking Algorithm Based on Improved RANSAC and KCF

In order to track the specific pedestrian targets in public places, a universal pedestrian tracking algorithm based on improved RANSAC and KCF (Kernel Correlation Filter) was proposed. First, the improved RANSAC algorithm is used to match the SURF descriptor, the surveillance video of the target is automatically identified and the location of the target is marked. Second, the KCF classifier and the foreground-background model based on the color feature are trained respectively to obtain the target confidence map through feature fusion, which are used to predict the location of the next frame of the target. Third, the improved algorithm and the original KCF were tracked on several public data sets respectively, the results show that accuracy of the improved algorithm is higher than the original one. Finally, the target tracking experiments of different public places are carried out, and the versatility of the proposed algorithm are verified.