Content Based Image Retrieval Using Color, Shape and Texture

Abstract

Our content based image retrieval (CBIR) system is capable of retrieving similar images from image database based on color, shape and texture. Color, texture and shape informations are primitive image descriptors and play a significant role in image retrieval purpose. CBIR systems extract these embedded image features from images and then utilize them to compare similarity between a query image and database images. The three fundamental bases in content-based image retrieval are features extraction, features matching, and retrieval of similar image for the matched features. In this work, we present a framework for the CBIR system providing color, texture and shape feature based image retrieval methods along with a combined retrieval method of all these three. In our CBIR system, as we observed, the retrieval efficiency is found to be satisfactory in terms of precision and recall.

Keywords: CBIR, Precision, Recall