A moving target detection method based on Dempster-Shafer Fusion

Moving target recognition has always been an important task in the field of computer vision. In this paper, we proposed a three-stage moving object detection method to solve the problem of difficulty in detecting target objects of the presence of complex backgrounds and occludes. This method first uses OTSU to perform threshold segmentation on the four color channels, then uses the Dempster Shafer theory to introduce historical information and fuse the segmentation results, and finally uses the dilation like method to compensate for the loss of information on the fusion process, identifying with the target object. This algorithm has been applied to the actual UAV formation flight task and has been proven to be stable and real-time.