Computer Engineering and Applications ›› 2015, Vol. 51 ›› Issue (22): 172-175.

Previous Articles     Next Articles

Edge histogram for sketch based image retrieval using nonsubsampled contourlet edge detection.Computer Engineering and Applications

CHEN Xi, XIAO Jian   

  1. Computer & Communication Engineering Institute, Changsha University of Science & Technology, Changsha 410114, China
  • Online:2015-11-15 Published:2015-11-16

基于NSCT边缘直方图的草图检索

陈  曦,肖  建   

  1. 长沙理工大学 计算机与通信工程学院,长沙 410114

Abstract: There is a wide range applications of Sketch Based Image Retrieval. Considering the differences of sketches and images, it is to be proposed in this paper that a new approach of Sketch Based Image Retrieval through using NonSubsampled Contourlet(NSCT) transform and Edge Histogram Descriptor. By?using?NSCT?transform?decomposition?of?image?to?do?edge
detection?for?all?subbands, and?extract?the?edge?histogram?of?the?subbands??that?is?used?as?describing?the features?of??images
and?sketches. After comparing the method to the edge histogram descriptor, the experimental results show that the retrieval results are more accurate.

Key words: Sketch Based Image Retrieval(SBIR), NonSubsampled ConTourlet(NSCT), edge histogram descriptor, edge detection

摘要: 基于草图的图像检索,存在着大量的应用空间。针对查询草图和数据库图像的差异,提出了一种利用非降采样Cotourlet(NSCT)变换和边缘直方图的草图检索算法。利用NSCT变换分解图像,对所有子带进行边缘检测,提取子带的边缘直方图用来作为描述图像和草图的特征。将该方法与边缘直方图比较实验表明,其检索效果更准确。

关键词: 草图检索, 非降采样Contourlet, 边缘直方图, 边缘检测