Content Based Document Image Indexing and Retrieval using Hybrid of Quadtree and Sparse Matrix Structure

Abstract

The fundamental consideration in the design & development of a content based image retrieval system is to extract the image features that best represent the image contents in a database. Currently several image indexing approaches based on Quadtree do exist. In this paper, the proposed approach is primarily concentrating on database classification for efficient image representation and effective retrieval using Quadtree decomposition technique augmented by sparse based index structure. In this proposed framework, the efficient storage of images is carried out in the beginning using Quadtree decomposition by routing the images in their respective subsets. This consequently enables us to directly retrieve the closely matching images from the appropriate subset(s) while discarding the other disparate. The fine retrieval process is eventually achieved with the help of already maintained sparse based information vector, prepared at the time of storage of images. Thus the proposed approach has the competence to significantly increase the accuracy in the image retrieval while simultaneously minimizing the computational complexity.