摘要
针对树结构矢量量化存在所选码字不一定是最佳匹配码字的问题,提出一种快速搜索算法,可以准确地找到输入矢量的最佳匹配码字.该方法利用了三角不等式作为继续搜索与停止搜索的依据,可以快速排除不可能节点,保留可能节点.实验结果表明,用该方法恢复的图像质量要好于双路搜索方法恢复的图像.
Aimed at solving the problem which consists in tree-structured vector quantization that the selected codeword is not sure of the closest one for the input vector. This paper presents a fast search algorithm, which can find the closest codeword. This algorithm utilizes triangle inequation to determine searching further or not, which can eliminate impossible node and reserve possible node quickly. Experiment result shows that the proposed algorithm does better than two-path search algorithm in recovery image.
出处
《电子器件》
CAS
2007年第3期1061-1063,共3页
Chinese Journal of Electron Devices
关键词
树结构
矢量量化
三角不等式
恢复图像
Tree-structured
Vector Quantization
Triangle inequation
Recovery image
