Image retrieval based on color tone variance difference feature

Interest is being concentrated on effective feature extraction methods for content based image retrieval. In this paper, we propose a RGB color retrieval technique based on color tone variance difference for image retrieval that is robust with respect to object size. After RGB images are arranged in color, RGB size order, color tone variance difference between colors RGB was used as the feature vector. Then each calculated variance difference was quantized into three areas and re-arranged. The proposed algorithm demonstrated improved performance compared to existing color histogram, CCV and color correlogram methods, through experiments.