Performance measure of color and texture in visual content retrieval in RGB color space

Abstract

Feature extraction simplifies the amount of information needed to describe the properties of an image accurately. This paper measures the performance of a CBIR system based on texture feature against combination of both color and texture feature. A Gray Level Co-occurrence Matrix is calculated for computing the texture feature of an image. Using these textual parameters similar images are extracted from a data set. RGB color space is considered for color feature extraction. Global Color Histogram is generated and calculated color features are represented as one dimensional feature vector. Then we combined both color and texture features to retrieve similar images from the dataset. In both situations Euclidean distance is used to measure the similarity of two images. By this experiment it is found that the system which uses the combination of color and texture has better performance in retrieving similar images from the dataset.