融合颜色和边缘特征的织物图像检索算法  

Fabric Image Retrieval Method of Blending the Color and Edge Feature

作　　者：崔红静 景军锋[1] 张缓缓[1] 苏泽斌 CUI Hongjing;JING Junfeng;ZHANG Huanhuan;SU Zebin(Xi′an Polytechnic University,Shaanxi Xi′an,710600)

机构地区：[1]西安工程大学

出　　处：《棉纺织技术》2019年第10期35-39,共5页Cotton Textile Technology

基　　金：国家自然科学基金(61301276);陕西省重点研发计划(2017GY-003);陕西省高校科协青年人才托举计划项目(20180115);陕西省教育厅科研计划项目资助(18JK0338)

摘　　要：为解决单一特征检索精度不高的问题,提出一种基于分块颜色直方图和边缘方向直方图的检索算法。将图像转换到HSV颜色空间,并进行图像分块得到子块颜色特征;通过Canny算子提取图像边缘信息,获得边缘方向特征;利用卡方距离分别度量颜色特征与边缘特征的相似性,实现图像检索。该算法对织物图像的平均检索查准率达80.83%。认为:该织物图像检索系统对织物检索有较高针对性,能够获得较好的检索效果。In order to solve the issue of lower retrieval accuracy on the single characteristic,a retrieval method based on block color histogram and edge direction histogram was put forward.The image was transferred to HSV color space and the color characteristics of subblock were obtained by separating the image block.The edge information of the image was extracted by Canny edge detector and the edge direction characteristics were obtained.The similarity of color characteristics and edge characteristics were measured by Chi-square distance respectively.The image retrieval was achieved.Precision ratio of mean retrieval of the algorithm on the fabric image was reached 80.83%.It is considered that the fabric image retrieval system has higher pertinence on fabric retrieval.Better retrieval effect can be obtained.

关 键 词：织物图像检索 分块颜色特征 边缘特征 HSV颜色空间 查准率 查全率 

分 类 号：TS101[轻工技术与工程—纺织工程] TP391.3[轻工技术与工程—纺织科学与工程]