The Algorithm of Micro Moving Target Detection in the Video Images

In order to effectively replace the background of micro moving targets in the video sequences, it is important to solve the problem of micro moving targets extraction. We propose a new algorithm of micro moving target detection in video images. Firstly, Adaboost was used to train skin samples to obtain the skin color data and determine the threshold range of skin color. Secondly, in order to obtain the rough binary images, we fused the binary images of skin detection with the binary images of micro targets by the Otsu. Thirdly, morphological processing and the holes filling processing were used to overcome the rough binary images, in order to obtain the complete binary templates. Lastly, the algorithm successfully accomplished the background replacement of the video sequences. The experiments demonstrate that the proposed algorithm in this paper is effective, which can segment micro moving targets accurately and complete background replacement.