Published Paper Details:

Abstract : In this paper, color histograms, texture using wavelet transform and shape feature using thresholding is used to retrieve images from the database .The database includes 500 images of five different categories. Moreover success of this content based image retrieval depends on the method used to create feature vector and relativity measure, the accuracy of wavelet transform engaged. Retrieving images can be increases the accuracy of the system based on color, feature and shape relativity. For relativity measure between the database images and query image Mahalanobis, Minkowiski, Chebyshev, Cosine Relativity, Spearman correlation are used ,at the end aggregation of mean and relative deviation is done and aggregation is then arranged in increasing order i.e. minimum to maximum. This Arranging order shows images having minimum values are more accurate retrieved image. The Proposed system can better the retrieval accuracy that has been shown by the experimental results. The system is tested on 500 images of five different categories downloaded from the internet