ABSTRACT

We see the advancement in image acquisition technologies and storage systems which always encourages us to design a sophisticated system to retrieve the images effectively. In this paper, we describe the novel approach for image retrieval based on image contents. It involves creation of feature database for all database images which includes formation of feature vector by applying the two methods one is by applying Kekre’s transform over row and column vectors and secondly by applying Kekre’s transform over row - column variance vectors of image. Further we apply a similarity measure to compare the query image and the database images. Finally we retrieve similar images from database based on the pre determined threshold. The database of 525 images of seven different categories (75 from each category) is used for demonstration to compare the performance of these algorithms using precision and recall as parameters.