Color and texture features for content based image retrieval system

Abstract

Content Based Image Retrieval System retrieves images using color, texture and shape properties of the image. Different methods which are implemented in this paper are Discrete wavelet transform (DWT), Gabor wavelet transform (GWT), Color histogram (CH), color autocorrelogram (CA). Integration of color and texture features is done using different methods and their comparison is done using precision and recall as performance measures. .DWT (D) method is implemented using the combination of statistical mean and standard deviation features and perceptual feature directionality. The best results are obtained with GWT (A) + CH method as compared to all other ten methods as phase information from the Gabor transformed coefficients is taken into consideration.