基于改进的边缘直方图的图像检索方法  

Methods of Image Retrieval Based on Improved Edge Histogram

作　　者：任平红[1] 陈矗[1] 

机构地区：[1]曲阜师范大学(日照校区)计算机科学学院,山东日照276826

出　　处：《计算机技术与发展》2007年第8期183-186,共4页Computer Technology and Development

基　　金：山东省优秀中青年科学家奖励基金(2005BS01016);曲阜师范大学科研启动基金

摘　　要：利用边缘直方图检索图像,因为人眼并不敏感的细微边缘也参与匹配,所以会影响检索效果。文中提出了一种改进的基于边缘直方图的检索方法,通过阈值把图像分割成目标和背景,利用梯度算子提取目标和背景之间的主要的边缘信息,避免了目标内部或背景内部的细微边缘的影响,更符合人眼的视觉特性。实验结果表明,其检索效果优于基于边缘直方图的方法。Because people＇ s vision is not sensitive to the slight edge that participated in matching, making use of gray edge histogram to retrieval image has an effect on the result. An approach to retrieve image based on improved gray edge histogram is proposed in this paper. According to image segmentation based on threshold value, the image has been divided into target and background. Only the main information of edge between target and background is extracted by means of gradient operator. Without the adverse effect of the edge inside target or background, so this approach is more suitable for visual characteristic. Experiment results improved that this improved approach is better than that based on gray edge histogram.

关 键 词：边缘直方图 边缘检测 灰度阈值 图像分割 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]