The H.264 standard is a new state-of-the-art video coding standard with extensive applications. This paper presents a
simple and efficient approach for motion segmentation in H.264 compressed video. Several preprocessing steps are used
before actual motion segmentation. The raw motion vector (MV) field extracted from H.264 video is first spatially
normalized and then accumulated by the forward projection scheme to obtain the dense MV field. The following global
motion compensation is performed on the accumulated MV field to acquire the residual MV field. Based on the residual
MV field, a hybrid scheme including edge detection and region growing for motion segmentation is proposed. The edge
map is used as a mask to guide region growing, which is created by Canny operator based on the magnitude map of
residual MV field. At last, hypothesis testing as the major postprocessing technique is exploited to distinguish between
the background and different moving objects. Experiment results demonstrate that the high-efficiency performance and
good segmentation quality of the proposed approach.
This paper proposes a multiresolution method for video object segmentation in the compression domain. We first calculate global motion parameters using only background macroblocks with tiny residual dc coefficients of the P frame, and then get true motion vectors projected to the immediate adjoined I frame. The basic layer image is obtained with only dc coefficients of the I frame. The enhancement texture characteristics are provided by the ac coefficients for partial decoding. The true object motion vectors and the basic layer image are fed into a morphological motion filter to get the lowest-resolution regions of moving objects, called the layer4 region of interest (L4-ROI). Only some of the ac coefficients in L4-ROI are decoded to obtain a higher-resolution image, called layer3, that mainly consists of blocks of the moving object. The moving object of interest in the highest resolution is obtained from a morphological motion filter with L2-ROI and the true object motion vectors. The number of ac coefficients determines the resulting resolution. Experiments show that the new algorithm can extract multiresolution moving objects efficiently.