Computer Engineering and Applications ›› 2017, Vol. 53 ›› Issue (6): 246-251.DOI: 10.3778/j.issn.1002-8331.1509-0086

Previous Articles     Next Articles

Local matching algorithm for product image search

WANG Qihao1,2, WANG Bin1,2   

  1. 1.Information Engineering College, Nanjing University of Finance & Economics, Nanjing 210046, China
    2.Collaborative Innovation Center for Modern Grain Circulation and Safety, Nanjing 210023, China
  • Online:2017-03-15 Published:2017-05-11


王其浩1,2,王  斌1,2   

  1. 1.南京财经大学 信息工程学院,南京 210046
    2.江苏省现代粮食流通与安全协同创新中心,南京 210023

Abstract: Image matching is one of the key issues in product image search. An algorithm for product image search based on local matching is proposed to solve the problem of lacking local matching in existing image shopping websites. This algorithm allows user-defined regions of interests and uses SIFT(Scale Invariant Feature Transform) descriptors to extract the visual features of these regions for matching. Experiment results demonstrate that this algorithm achieves ideal expected results, and better meets the actual shopping demands of the users.

Key words:  Scale Invariant Feature Transform(SIFT), image matching, image retrieval, local features

摘要: 图像匹配是图像购物搜索的一个关键问题。针对现有的图像购物搜索网站利用底层特征无法进行局部匹配的问题,提出一种基于局部匹配的图像购物搜索方法,该方法允许用户自定义感兴趣区域,并利用SIFT(Scale Invariant Feature Transform)描述子提取该区域的视觉特征进行匹配。实验结果表明,该方法取得了较理想的预期效果,更好地满足了用户的实际购物搜索需求。

关键词: 尺度不变特征变换(SIFT), 图像匹配, 图像搜索, 局部特征