摘要
提出了一种直接从压缩域提取显著闭合边界进行图像检索的算法。从压缩域直接提取块边缘,使用基于网络流理论的比率轮廓算法从块边缘图像中提取显著闭合边界,用傅立叶描述子刻画显著闭合边界,进行图像检索。实验结果表明,直接从压缩域提取显著闭合边界,无需解压,减少了计算量,提取的显著闭合边界应用在图像检索中取得了较高的准确率。
A new image retrieval algorithm which based on salient closed boundary extracted directly from compressed domain is proposed, Firstly, block edges is extracted directly from compressed domain without decompressing. Secondly, network flow based ratio contour algorithm is used to extract salient closed boundary, Finally, Fourier descriptors is used to describe the closed boundary and retrieve images. Experimental results demonstrate that extracting block edge from compressed domain saves a lot of computation cost, and using the detected closed boundary in the image retrieval gets a high precision.
出处
《计算机工程与设计》
CSCD
北大核心
2008年第9期2286-2289,2292,共5页
Computer Engineering and Design
基金
重庆市科委自然科学基金项目(CSTC-2006BB2309)
