Statistical Distribution-based Color Image Retrival
Karthika Murugandi Reddiar Seetharaman, Woothukadu Thirumaran Chembian · 2017
This paper proposes a novel technique, based on distributional approaches with distance measure, i.e. Chernoff distance measure. Since the proposed system is an automatic image retrieval, it is difficult to understand the nature of the query image and which distribution it follows. The Chernoff distance measure overcome this problem, because it adapts itself according to the nature of the images, viz. the Chernoff distance could be adopted though the query image is not distributed to Gaussian or mixed or even if distribution free. This is the main advantage of the proposed technique. In order to examine the proposed technique, an image database is constructed, which contains manifold images such as texture, structure, blurred, noise, artifacts images and their features.