A Study on the Very Low bit-rate Video Coding Based on Isodensity areas

点击翻译按钮获取中文摘要

In MPEG4, not only promotion of a low bit rate but also an object scalability depended on the contents of images are strongly requested. In this research project, an efficient video coding algorithm for the upper half of human body is propsed. In this coding system, a facial area, which is a moving object in image sequence, is extracted from both sending and receiving side without any control information. The candidate areas of person are segmented using color information and the moving areas. The color information, which represent human skin color, is extracted with modified HSV color system. The moving areas are estimated using the motion parameters of objects which are utilized for constructing a previous frame. Then a human face is extracted by merging the candidate areas based on hue-element of modified HSV color system. For a computer simulation, the standard video "Claire" and "Miss America" are used and facial areas are extracted precisely in all frames. The experimental results of "Claire" show that a signal to noise ratio is 34.1dB at 13.4kbits/sec and "Miss America" show that a signal to noise ratio is 33.1dB at 17.4kbits/sec on average respectively. From these experimental results, the prospects of adopting this method for a coding system of MPEG4 can be seriously considered.