基于内容的图像检索在电子商务中的应用

摘要

This paper applied content based image retrieval into electronic commerce, analyzed the situation and advantage of this application, and explored several key issues. Fist the product image was divided into 3 annular regions, and redundant information of the outermost annular was removed, then main color features of core area and center area was used into firstly retrieval, and the texture features of these areas into secondary retrieval, when retrieval different weighting factor was applied into features of core area and center area according to the importance of different area. The result indicates that this algorithm can bring better retrieval effect for some categories of product image.%将基于内容的图像检索应用于电子商务，并分析了这项应用的优势和现状，探索研究了基于内容的图像检索应用于电子商务的几个关键问题．提出将商品图像分为3个环状区域，先去除商品图像最外层环状区域冗余信息，然后利用商品图像核心区域和中心区域的主颜色特征进行检索，再提取商品图像核心区域和中心区域的纹理特征进行二次检索．检索时根据图像核心区域的重要性，对图像核心区域和中心区域的特征采用不同的加权系数．经实验验证，本文采用的算法对一些类别商品图像有较好的检索效果．