摘要
现有边缘匹配矢量量化(SMVQ)算法没有考虑码字索引分布规律。针对该问题,通过理论分析得到码字索引严重非均匀分布的规律,基于该规律提出结合哈夫曼编码的SMVQ图像压缩算法。实验结果表明,与穷尽搜索算法相比,该算法在保持相同编码质量的情况下,编码码率下降50%,与SMVQ相比,其PSNR提高0.8dB,编码码率下降30%。
Aiming at the problem that existing Side-Match Vector Quantization(SMVQ) algorithms do not consider the distribution rule of codeword index, this paper obtain the rule of serious non-uniform distribution through theoretical analysis. A SMVQ image compression algorithm combined with Huffman coding is proposed by using the rule. Experimental results show that compared with full searching algorithm, the coding rate of this algorithm decreases by 50% while having equal coding quality. Compared with SMVQ, its PNSR gets a raise of 0.8dB while coding rate decreases by 30%.
出处
《计算机工程》
CAS
CSCD
北大核心
2010年第4期218-219,222,共3页
Computer Engineering
基金
国家自然科学基金资助项目(60672041)
国家"863"计划基金资助项目(E20050605)
关键词
矢量量化
边缘匹配
哈夫曼编码
图像压缩
Vector Quantization(VQ)
side-match
Huffman coding
image compression
作者简介
邓宏贵(1965-),男,教授、博士后,主研方向:智能优化算法,图像处理
郭晟伟,硕士研究生.E-mall:guoshengwei2000@163.com
李志坚,硕士研究生