A moving target detection algorithm based on dynamic scenes

Abstract

Due to the impact caused by background, shadow, change of illumination, moving speed and so on, detection of moving target fails easily with its anti-noise performance no well. In accordance with these problems, this paper proposes a novel moving target detection algorithm based on dynamic scenes. By the combination of five-image difference algorithm and background subtraction algorithm, a complete outline of moving target can be obtained after preprocessed by median filtering method. Moreover, through making use of rectangular contour model, the cast shadow of a moving target can be effectively removed. The results show that the algorithm can extract moving target from dynamic scenes more accurately and completely, with real-time guaranteed.