Query by Images: An Electronic Apparel Catalog Retrieval System

H. Yuan, F. Wang, and X.-P. Zhang (Canada)

Keywords

contentbased image retrieval (CBIR), Gaussian mixture model (GMM), wavelet domain, electronic commerce, electronic apparel catalog retrieval system (EACRS)

Abstract

Content-based image retrieval (CBIR) methods, in which image texture and color features are used to index and retrieve images, are valuable for online shopping, but have not been widely employed in e-commerce applications. In this paper, we present a CBIR based electronic apparel catalog retrieval system (EACRS), which can automatically retrieve product images similar to query images according to certain user requirements. The experiments conducted on a hat image database demonstrate that the proposed system, EACRS, is effective in retrieving similar apparel product images based on the combination of the texture and color features and can be further enhanced when relevance feedback is introduced.

Important Links:



Go Back