基于边缘检测与颜色目标定位的图像检索

Image Retrieval Based on Edge Detection and Goal Location

提出一种先利用图像边缘信息缩小检索范围。再利用图像颜色信息进行目标定位的图像检索算法。首先。提取图像边缘信息。利用图像位错率进行边缘信息比较，位错率越小图像边缘信息越接近；然后，利用主色调确定图像的目标块，进行目标子块间的比较。算法忽略图像中不重要的背景因素，有效利用重要的目标因素。从而提高了检索效率。仿真实验证明该算法比单一使用边缘检测、颜色目标定位具有更好的检索性能。

In this paper, a new method was introduced in the Content Based Image Retrieval that located the goal with color after detecting the image edge. The arithmetic compared image edge by BER （bit error rate）. It was considered that the edge was more alike when the distance was smaller between the two images. And the method compared the goal of the image in main color. The subordinate background factor was ignored, and the main goal was effectively utilized, which improved the efficiency of the image retrieval. Experiments show that the combine of the edge feature and color feature can more improve the retrieval performance than single one.