Dual frame differences based background extraction algorithm

Abstract

Background subtraction is very important part of surveillance applications for successful segmentation of objects from video sequences. The robust initial background extraction is crucial in any background subtraction. In this paper, we propose an algorithm to extract initial background from surveillance videos using dual frame differences and morphological processing. With the proposed algorithm, the initial background can be extracted accurately and quickly. Experimental results for various environmental sequences are provided to demonstrate the robustness, accuracy and effectiveness of our method.