An algorithm for video monitoring under a slow moving background

Abstract

In this paper, a video monitoring algorithm under a slow moving background is proposed. An affine transformation model is used to describe the background image movement and two methods are given to find the affine transformation model parameters. The affine transformation is used to check the matching performance of the two frames and subsequently a subtraction operation based on block difference is performed for scene change detection. In order to finalize the detecting result, a series of image processing operations, including the adaptive threshold, morphological dilation and erosion operation, and region labeling have to be performed. Several experiments are given to show that the proposed algorithm is efficient.