An Improved CamShift Algorithm Based on FAST-SIFT Feature Detection Matching

The tracking accuracy of the traditional CamShift moving target tracking algorithm will drastically reduce when the target is occluded or the color between the target and background is similar, moreover, there may even be serious problems with missing target. To solve the problem, this paper proposes an improved algorithm based on FAST-SIFT feature detection matching. The algorithm firstly needs to manually select the tracking target as the target template, and then compares the target search window obtained by the traditional CamShift algorithm with the target template for color histogram comparison. If the Bhattacharyya coefficient between the target candidate area and the target template is greater than the detection threshold, the target tracking fails. Then, the FAST-SIFT feature detection matching algorithm is used to match the previously obtained target template with the current frame image, which re-determines the position information of the lost target. Therefore, the target can be further tracked by the traditional CamShift algorithm. Simulation experiments show that the improved algorithm can still guarantee a good tracking effect in a relatively complicated background environment.