分形图像编码的快速细粒度迭代解码

该文基分形图像编码的固有特征，阐述和研究了分形图像编码的迭代解码方法的“可分级性”概念，在这一思想指导下，在迭代解码过程中用单缓冲算法取代传统分形编码中的双缓冲算法，在节省了内存空间开销的同时有交地实现了更细粒度上的质量可分级性解码，继而提出了一种块排序解码算法用于基于尺寸自适应块分割的分形编解码方案中，获得了更快的解码收敛速度，通过以上一系列算法深入地阐述了分形偏解码中“质量连续可分级性”的思想，给出了一个初步的实现模型并对结果作了详尽的分析。

Fast Fine Granularity Decoding of Fractal Image Coding

One of the most important characteristics of fractal image coding is its asymmetrical property of encoding and decoding processing. Coding time is rather long for domain codebook generation and domain-range matching operation, while decoding algorithm is relatively simple and fast. This makes it a good candidate for those storage and retrieval applications where coding processing is performed only once but decoding processing needs doing many times. "Resolution independence" has been claimed as one of the main advantages of fractal image compression. To a data stream produced by an encoder of fractal image coding, decoding can be done in different spatial resolution. A fast pyramidal decoding algorithm can gradually produce visual-satisfying version in many different spatial resolution even when the resolution is much larger than the original image. Reconstruction of the compressed image is achieved by computing the fixed point of the image transform T from its encoded coefficients using iterated function system. Every iterating of the image is an approach to final attractor. From this point of view, iterating process actually also gives a quality scalable decoding. Speeding up convergence of the decoding algorithm is an important research branch of fractal coding. Some algorithm aiming at it are fast mapping updating, ordered iteration, combination with VQ, etc.The work of this paper describes iterative algorithm in fractal decoding in a deeper sense. Double buffer mechanism is commonly used in nowadays decoding algorithms. However, it is shown in this paper that a single buffer algorithm is much more advisable in terms of computation and memory saving. With single buffer algorithm, the fine granularity iterative decoding in which a single mapping is seemed as a minimal iteration unit is easily accomplished. Additionally, an ordered decoding algorithm is proposed and used in decoders of the fractal coding schemes based on variable size block segmentation, resulting in a faster convergence speed. With the aforementioned single buffer mechanism and block ordered decoding algorithm, a fast fine granularity decoding scheme of fractal image coding is realized.