摘要
针对分形域上的基于内容的图像检索(content-based image retrieval,CBIR),提出了一种新颖的基于无搜索的自适应四叉树分割的快速分形编码方法,来提取图像特征,从而使图像检索的编码阶段速度显著提高。对一幅256×256图像编码,算法平均约需0.0485s,比何的方法约快70倍,并且解码图像质量良好。改进了匹配算法来实现图像的快速检索,其准确性要高于洪的方法;最后通过对分形码距及分形码块数的分析,提出了进一步提高检索速度的方案。
For the content-based image retrieval （CBIR） in fractal domain, a fast fractal encoding method is proposed to extract image features, which is based on a novel no-search and adaptive quadtree division. As a result, the fractal coding speed is significantly improved, it only needs 0.048 5 s on average for a 256×256 image and is approximately 70 times faster than He＇s algorithm besides better reconstructed image quality. Furthermore, image matching Hong＇s algorithm is also improved, enhanced the query accuracy consequently. In addition, a method to further accelerate image retrieval is presented based on the analysis to the distance and number of the fraetal codes.
出处
《计算机科学与探索》
CSCD
2009年第4期423-432,共10页
Journal of Frontiers of Computer Science and Technology
基金
国家自然科学基金No.60573172
高等学校博士点专项科研基金No.20070141014
辽宁省教育厅高等学校科学技术研究项目No.20040081~~
关键词
分形编码
无搜索
平均块
基于内容的图像检索
fractal coding
non-searching
average block
content-based image retrieval
