摘要
本文提出的基于纹理的图像检索算法RAH是以图像的能量谱直方图为基础,包括以下两个方面:第一,计算图像能量谱的半径和角度直方图的方法以及基于图像能量直方图的频率去提取纹理特征的算法;第二,基于纹理特征的相似度检测方程。这个实验结果表明,RAH算法优于一般的灰度共生矩阵纹理检索算法,具有较好的检索效果,比较适用于基于内容的图像检索。
A texture based image retrieval algorithm RAH based on the histogram of image power spectral in frequen- cy domain is proposed, which mainly includes the next two parts: first, the method of computing radius histogram and angular histogram and the algorithm of texture feature extraction based on the histogram of image power spectral in frequency domain, second, the similarity measurement function based on the texture feature. The experimental re- sults indicate that RAH is preferable to the traditional co-occurrence matrix based algorithm.
出处
《计算机科学》
CSCD
北大核心
2005年第1期194-197,201,共5页
Computer Science
基金
江苏省应用基础基金BJ20009