Abstract: This paper researches the image retrieval based on salient regions. It improves the extraction method of salient point of wavelet domain, solves the problems that content of image feature extraction is limited and the image content can not be effectively expressed. It presents the method for extracting salient regions by using salient points as clues. It can distinguish the background image from object image. For object image, it only retrieves the salient regions. It can effectively reduce the interference of background area in retrieval process, and improve the speed and accuracy of retrieval.

Key words: image retrieval, salient regions, wavelet transform, salient point

摘要： 研究基于显著区域的图像检索，对小波域显著点提取方法进行改进，解决提取图像特征内容有限、不能有效表达图像内容的问题。提出由显著点作为线索提取显著区域的方法，该方法能实现背景图与对象图的区分。对于对象图，仅对显著区域进行检索，有效减少图像检索过程中背景区域的干扰，提高了检索的速度和准确率。

关键词: 图像检索, 显著区域, 小波变换, 显著点

CLC Number: 

• TP391