Mohsen Azararjmand - Qazvin Branch, Islamic Azad University

With increasing size and volume of digital images, the need for content-based image retrieval has increased. This paper introduces a graph-based approach for contentimage retrieval. In the proposed method, an image presented by a set of regions and when dealing with image matchingproblems, we use graphs to represent images, this causedtransform the region correspondence estimation problem into an graph matching problem and define the image distance interms of the estimated graph correspondence . Then the difference between images obtained. Experimental resultsthat the proposed graph-theoretic image matching is acceptable.

Content Based Image retrieval, graph matching, image segmentation, matching matrix