机构地区: 浙江大学信息与电子工程学系
出 处: 《光电工程》 2006年第2期85-89,共5页
摘 要: Ray-space数据特征分析表明,其片内和片间均存在很大相关性,并且其数据有明显的方向性特征。通过最优宏块分割、不同精度的块匹配和多参考片预测结果分析,讨论了适合Ray-space数据特点的片内和片间预测技术。与传统的只利用片内相关性的压缩技术相比,利用上述两类相关性,并结合Ray-space数据方向性特征,能进一步提高编码压缩效率。在获得相同重建图像质量下,可使基于H.264的Ray-space压缩方法比基于H.263的方法在码率上节省50%,比JPEG的方法节省75%左右。 Analysis of characteristics of Ray-space indicates that there are high intra correlations within a slice of Ray-space, as well as inter correlations between adjacent slices, and significant directionality characteristic in Ray-space data. Through analyzing optimal macroblock segmentation, block matching with different precision and multi-slices prediction, the techniques of intra-slice and inter slice predications suited to Ray-space compression are discussed. Compared with traditional Ray-space coding techniques only utilizing intra-slice correlations, the Ray-space coding efficiency can be further improved if the two correlations are used combined with the directionality characteristic of Ray-space. In addition, compared with Ray-space compression method based on H.263, Ray-space compression method based H.264 can save 50% bit rate at the same reconstruction quality of Ray-space, and the economization is about 75% in comparison with JPEG.