基于内存优化和启发式深度优先搜索的小波零树图像编码算法

本文提出了一种基于内存优化和启发式深度优先搜索的小波零树图像编码算法,采用标志位和启发式的深度优先搜索策略完成零树编码中的排序和细化过程.采用一张标志位图标识细化过程中的重要系数,代替零树编码中的链表,节省了内存占用.同时采用启发式的深度优先搜索策略检索树结构中的不重要象素集合信息,不仅省去了存储排序全局信息的链表和位图,符合优化内存的需要,还减少了排序过程中的定位信息,提高了信噪比.实验表明,和现有的小波零树编码算法相比,较大地节省了编解码内存,信噪比虽略低于SPIHT(Set Partition In Hierarchal Tree)算法,但高于EZW(Embedded Zerotree Wavelet)、LZC(Listless Zerotree Coding)等编码算法,为硬件实现小波零树编码算法探讨了一条有效途径.

A Wavelet Zerotree Image Coding Method Based on Memory Optimization and Heuristic Depth Finding Strategy

A wavelet zerotree image coding method based on memory optimization and heurist ic Depth Finding Strategy (DFS) is presented in this paper.The form of flag and heuristic DFS are introduced in the sorting and refining process.The flag map in stead of lists in the refining process is used to keep significant wavelet coeff icients.The DFS is used to search insignificant wavelet sets in the sorting proc ess.Therefore,both the memory of coding and the oriented information of sorting a re decreased.Experimental results show that compared to the existing wavelet ze rotree coding method,the memory is significantly reduced.Although the PSNR is li ttle lower than SPIHT's,it is higher than EZW's and LZC's.This proposed method p rovides an efficient hardware implementation.