Automatic Moving Object Extraction in MPEG Video

Abstract

In this paper, we propose a moving object extraction
technique for MPEG coded data directly. It is a change based motion object extraction approach, which discriminates background and
moving objects by means of the higher-order statistics (HOS) performed on the inter frame differences of DC image. The DC image is partly decoded picture from the compressed video for the rapid reconstruction of image data. In order to employ an optimal threshold in moving object detection stage, the background is
detected by the Moment-preserving thresholding technique for each frame. Based on the background statistic, the proportion of background variance is employed to extract the final object mask by
comparison the fourth moment measure and the variance.
Experimental results have demonstrated that the proposed approach worked efficiently and shown a robust result for object extraction in compressed video.