Robust Background Acquisition from Dynamic Scene caused by a Moving Camera

Abstract

A background is a part that does not vary too much or frequently change in an image sequence. Using this assumption, it is presented a background reconstruction algorithm for not only static but also dynamic view in this paper. For generating background, we detect a region that has high correlation coefficient compared within prior pyramid image from the current image. Using detected regions, we calculate movement vector for each region and classify clusters of each pixel belonged in the region based on pixel intensities. Finally we eliminate clusters which have lower weight than threshold, and combine remained clusters for each pixel to generate multiple background clusters. Experimental results show that automatically detected background under camera moving.