从研究分形图像编码和零树编码各自的优劣点以及它们之间的结合点出发,寻找了一种基于图像纹理分析的分形和SPIHT混合编码,目的在于充分利用景物特征和人眼的视觉特性,提高分形变换和零树在图像编码领域内的协同能力。该方法利用SPIHT算法位平面编码的渐进特性,用基于灰度模型的统计特征分类方法将分形和SPIHT相结合,得到更符合人眼视觉特性的编码方案。实验结果表明,该方法能进行较好的图像块分类,并取得较高的压缩比,而且在人眼视觉允许的范围内且同时又要求高压缩比的情况下具有优势。

Starting off the research of respective advantages and disadvantages of fractal image coding and zerotree image coding and their bonding points, a hybrid image coding based on image texture analysis of fractal and SPIHT(Set Partition in Hierarchical Trees) is presented to make full use of scenery characteristics and eye's vision features and then improve the cooperation ability of fractal and zerotree in the area of image coding.The prominent feature of this method is that it makes full use of the progressiveness of bit plane coding used in SPIHT coder and combine the fractal and SPIHT with statistical characteristic taxonomy based on grey-model and then get the coding scheme that is more suitable for eye's vision features.The result shows this method can classify the image block clearly and get high compression rate.And this method gains a better effect when generally considers the recovery image quality and compression rate,under the condition of the permission of eyes and high compression rate.