出版社：Academy & Industry Research Collaboration Center (AIRCC)
摘要：In a content-based image retrieval system (CBIR), the main issue is to extract the imagefeatures that effectively represent the image contents in a database. Such an extraction requiresa detailed evaluation of retrieval performance of image features. This paper presents a reviewof fundamental aspects of content based image retrieval including feature extraction of colorand texture features. Commonly used color features including color moments, color histogramand color correlogram and Gabor texture are compared. The paper reviews the increase inefficiency of image retrieval when the color and texture features are combined. The similaritymeasures based on which matches are made and images are retrieved are also discussed. Thepaper discusses effective indexing and fast searching of images based on visual features.
关键词：CBIR; Color moments; Color histogram; Color correlogram; Gabor filter; Precision; Recall.
