New video segmentation algorithm using a combined segmentation measure forcontent-based coding

A new video segmentation algorithm using a combined segmentation measure fo r content-based coding is presented. The combined segmentation measure is a weighted sum of intensity, motion, and a change segmentation measure. The change segmentation measure is defined from a change image, which is obtain ed from the proposed change detector. The proposed change detector performs morphological erosion filtering to eliminate many inaccurate components, i ncluded in the resulting image obtained from a conventional change detector . Since the change segmentation measure is defined as an absolute change va lue difference between a pixel and its neighboring region, it can be an eff icient segmentation measure for the accurate segmentation of neighboring mo ving objects and static background regions. Therefore, the proposed combine d segmentation measure can determine exact boundaries even though the estim ated motion vectors around the boundaries of moving objects and static back ground regions are inaccurate and the intensities around the boundaries are similar. From the experimental results for a video segmentation, since the combined segmentation measure can accurately segment the boundaries betwee n moving objects and static background regions, the motion compensated imag es produced by the proposed segmentation algorithm show improved image qual ity compared to the conventional segmentation. (C) 2000 Society of Photo-Op tical Instrumentation Engineers. [S0091-3286(00)01208-3].