中文摘要：

充分利用深度图序列相邻帧间的局部相似性，并结合H．264帧内预测编码算法，提出了一种适用于深度图序列编码的新方法，即深度图序列的I帧编码采用H．264帧内预测结构而P帧编码采用基于分形理论的压缩方法。首先介绍了分形编码的理论基础，然后通过改进非均匀多层次六边形格点整像素运动搜索（UMHexagonS）方法，提出了一种更适用于分形编码的基于运动矢量场的自适应六边形搜索算法，从而形成了基于分形的深度图P帧预测编码结构。实验结果表明：本文提出的编码系统在很大程度上提高了H．264的编码性能，峰值信噪比差值（BDPSNR）平均提高了3．54dB，编码比特率差值（BDBR）平均降低40．98％，同时编码复杂度也大大降低，平均编码时间减少了85．27％。

英文摘要：

Depth map compression is the crucial technology in three dimentional video （3DV） coding sys- tem. To obtain high compression efficiency and low encoding complexity for depth map-sequence,this pa- per proposes a novel coding algorithm for depth map-sequence based on fractal. This applicable scheme for depth map-sequence coding is proposed by adequately exploiting local similarity between adjacent frames in depth map-sequence and combining intra-prediction method in H. 264. In the proposed scheme, the I frame in depth map-sequence is encoded by adopting intra-prediction structure in H. 264, while the P frame is encoded by fractal compression method. Firstly,the theoretical basis of fractal coding is pres- ented. Then,a new motion-vector-field-based adaptive hexagon search algorithm, which is applicable to fractal coding, is proposed by improving unsymmetrical multi-level hexagon search （UMHexagonS） method. Finally, the P frame prediction structure based on fractal in depth map-sequence coding is formed. Experimental results indicate that the proposed coding system can improve the overall perform- ance of the H. 264 coding system with an average Bjoontegaard delta peak signal-to-noise ratio （BDPSNR） gain of 3.54 dB and an average Bjoontegaard delta bit rate （BDBR） reduction of 40.98%. Furthermore, the computational complexity is reduced and the entire encoding time on average is saved by 85.27 %. In addition, the proposed algorithm shows certain advantages compared with other interna- tional representation methods in this field.