Canny特征在基于内容的图像检索中的应用

A content-based image retrieval method based on Canny feature

在基于内容的图像检索中,枪支一类图像的检索有其自身难点.将Canny边缘检测算法应用于图像检索中,以待检图像的Canny特征作为检索特征,提出了一种结合矩阵相似度理论的基于内容的图像检索算法.并以手枪图像的检索为例,对算法进行了实验测试.实验结果表明,该图像检索算法具有较高的查准率、查全率和较低的误检率.

Duo to the unique appearance of the gun image, most content - based image retrieval method is not able to retrieve gun image effectively. By applying canny algorithm, a content -based image retrieval algo- rithm based on the similarity of matrices is designed in this article. By experiment, the algorithm to the retrieval of gun image is applied. The results have testified the great performance of the algorithm in terms of a high recall ratio, precision ratio and low false alarm ratio in the retrieval.