Effective Content Based Image Retrieval techniques using clustering for multiple features

Abstract

Content based Image Retrieval System overcomes the drawback of traditional text based image retrieval systems because of large image database, it is much time consuming and labour intensive. In CBIR system search will analyze these image contents rather than metadata of images as keywords and other descriptions related to image. Purpose of this paper is to include more visual features simultaneously rather than using them individually. Here we implement a series of algorithms for combining colour and texture features extraction in CBIR for more accurate results, colour feature vectors and clusters are formed using statistical colour moments along with hierarchical and K-means clustering technique and for texture feature extraction Gabor wavelets algorithm is applied and on retrieved images relevance feedback analysis is performed for automatic indexing of images for future.

Keywords: CBIR , RGB, Hierarchical, K-means, Centroid, Statistics distance, Gabor wavelet, Colour Moment