Un-Supervised, Semi-Supervised and Supervised Image Retrieval Based on Conceptual Features

Abstract

To effectively and efficiently retrieve desired images from a large image database, an intuitional and common type of approaches is text-based image retrieval which accesses the images by comparing conceptual terms of a query and image data. Unfortunately, this type of image retrieval is not easy to earn users' satisfactions due to the problem of the image database maintenance. Another useful type is content-based image retrieval which determines the most relevant images according to the content comparisons between a query image and searched images. Although the content-based image retrieval can avoid the problems of text-based image retrieval, it cannot certainly generate good results because of the semantic gap between low-level visual features and high-level concepts. Therefore, in this paper, we propose three types of content-based image retrieval and a conceptual feature to improve the quality of content-based image retrieval. The experimental results show that, the supervised image retrieval using the proposed conceptual features can bring out better retrieval results than the traditional image retrieval using low-level visual features.