摘要
在分析利用颜色直方图进行图像检索时存在问题的基础上,提出了一种新的图像空间描述符。首先利用颜色矢量角的特性将图像分为边缘区和平滑区两部分,然后,针对两种区域的不同分布特点,综合利用图像的边缘信息、形状信息及空间分布特性,采用两种不同的描述符——方向颜色矢量角直方图和空间联合分布熵分别来提取图像的边缘区和平滑区的分布特征。实验结果表明,将其用于图像检索,取得了很好的检索效果。
The content-based image retrieval on histogram was analyzed and a new spatial image descriptor for image retrieval was proposed based on the color vector angle. Firstly, the image was divided into two parts by the color vector angle： edge domain and smooth domain. Then, according to the different distribution of the two domain, two different descriptors- directional color vector angle histogram and spatial union-distribution entropy were proposed to extract the image features. The edge information, shape feature and spatial information were taken into account. Experimental results show that the proposed method not only has sound retrieval performance but also has the robustness against illumination changes, position translation, and scale translation of the image content.
出处
《光电工程》
EI
CAS
CSCD
北大核心
2007年第3期77-81,88,共6页
Opto-Electronic Engineering
关键词
图像检索
颜色矢量角
边缘检测
颜色直方图
Image retrieval
Color vector angle
Edge detection
Color histogram
