Moving object detection in dynamic background

Abstract

A new method of detecting moving object in dynamic background is proposed in this paper. At first, an adaptive threshold Harris algorithm is proposed in this paper to extract feature points, then, SIFT algorithm is used to describe these extracted feature points. The similarity function is used to match feature points and RANSAC algorithm is used to eliminate the pseudo matches. According to the correct matches, we get the affine transformation matrix which used to compensate the motion of background caused by camera motion, and update the dynamic background with the background model. Finally, the moving object can be detected by background subtraction method. Experimental results show that the method presented in this paper improves the accuracy of feature point extraction and detects moving target in dynamic background accurately.