Color perception Histogram for Image Retrieval using Multiple Similarity Measures

This study aims to increase the retrieval efficiency of proposed image retrieval system on the basis of color content. A new idea of feature extraction based on color perception histogram is proposed. First, the color histogram is constructed for HSV image. Secondly, the true color and grey color components are identified based on hue and intensity. The weight for true and grey color components is calculated using NBS distance. An updated histogram is constructed using weighted true and grey color values. The color features extracted from the updated histogram of query image and for all the images in image database are compared with existing color histogram based technique by using multiple similarity measures. Experimental results show that proposed image retrieval based on the color perception histogram gives higher retrieval performance in terms of high average precision and average recall with less computational complexity.