3D-color vector quantization for image retrieval systems (データベースシステム)

概要

論文の詳細を見る
In this paper we present a new approach for extracting color feature for image retrieval systems. The system introduces color feature extraction using histogram of Three Dimensional (3D)-Color Vector Quantization of RGB color space. The main idea of this approach is that the system uniformly represents image colors in certain positions of RGB's vector color space. It intends to reduce a complexity RGB colors in the image and unifies the close colors in the vector space. In advance, our system pre-proceeds noise removal to enhance the quality of an image before extracting the feature. We apply 3D-Color Vector Quantization in each block of image partition to keep a close similarity between image query and image database in the matching process for each block of partitioned image. We evaluated our system using COREL image collections containing different categories of images. The experimental results clarified effectiveness of the proposed system to improve the accuracy for image retrieval.
一般社団法人情報処理学会の論文
2008-09-14