【作者】 王忠欣； 熊璋； 蒲菊华；

【Author】 Wang Zhong-xin+, Xiong Zhang, Pu Ju-hua (School of Computer Science and Engineer, Beijing University of Aeronautics and Astronautics, Beijing 100083, China)

【机构】 北京航空航天大学计算机学院；

【摘要】 论文提出了一种基于MP(MatchingPursuits)变换的视频编码算法,该算法能对感兴趣区域优先编码,在码速率一定的情况下,提高视频的主观感受质量。该算法利用加权的能量块搜索策略以及向量比较的方法改进了MP变换的内积搜索过程,降低了算法的计算复杂度,并且输出码速率细粒度可调,具有较好的应用前景。更多还原

【Abstract】 When communicating at a very-low bit-rate, video coders are unable to preserve high visual quality for all images. In these images, some regions that user is interested in re-quire high temporal resolution. But most existing compression schemes treat the entire image uniformly. In order to achieve a better Perceptual Quality without compromising the frame rate, a selection of key regions according to human viewing and understanding may therefore be useful: it allows extracting regions of interest(ROI) and to code them with a higher quality, while the remainder of the image is coarsely transmitted. Such an approach requires a signal decomposition allowing an adaptive spatial variant bit allocation. Matching Pursuits (MP) is an iterative procedure leading to such a signal representation. Incorporated within a motion compensated video codec, MP allows for adaptive representation of the Displaced Frame Dif-ference. Bit allocation procedures for Matching Pursuits differ from the ones used for DCT or other transform-based signal representations. For MP, bit allocation is equivalent to function selection. This makes Matching Pursuits well-suited for spatial variant bit allocation, since functions matched to important objects can be chosen. In this paper, a weighted energy search strategy is applied to find coded atoms. By this way, Matching Pursuits can be able to allocate an available bit-budget according to a semantic understanding of the scene. This se-mantic information can be provided either automatically or interactively. For simplicity, the algorithm is extended to use a dynamic region of interest defined by face location using OpenCV. Second, although the matching pursuit is known to be very effective in coding mo-tion-compensated residuals, its massive computation required for finding atoms hinders its use from practical application. In this paper, we present a very simple and intuitive idea al-lowing one to preclude substantial number of basis functions from the inner product search based on simple comparison of vector norm. By this the computational load is reduced dras-tically. The proposed method does not change the number of basis functions, nor the search size. So our novel fast matching pursuit method yields exactly the same performance as the regular matching pursuit. Experimental results show the algorithm has a better Perceptual Quality on a given bit-rate and the number of inner product calculation is only about 30% of the conventional separability-based fast method.更多还原