Hierarchical segmentation methods for region-based video coding

Extraction of stationary objects, including the background, and restriction of the coding to moving objects, has been widely proposed in the literature as a means of achieving very low bit rate transmission, while ensuring high image quality. In cases where it is not feasible to code and transmit all the objects identified as moving, due to bandwidth limitations, we should ensure that at least the most visually important objects (i.e. the ones that contribute most to the quality improvement) are coded. In this paper, different methods for segmenting image sequences and obtaining the moving objects are investigated. For each segmentation method, we identify a hierarchical ranking of the moving objects, in terms of visual importance. We study the influence of the segmentation algorithm's choice on the performance of a region-based video coding scheme, and we present simulation results that demonstrate the efficiency of the criteria proposed for hierarchical moving object coding.