Feature based image retrieval of images for CBIR
Pradnya Rane, Pallavi Kulkarni, Suchita Patil, B. B. Meshram · 2011
Extensive use of images in various applications gives immense importance to Content based image retrieval. There are many ways to retrieve image from image database depending upon various image features like DCT coefficients, mean, median entropy etc. Histogram [6] is very simple and old approach for image retrieval. This paper describes a combined approach to content based image retrieval considering DCT- coefficients of R,G,B planes separately and basic image features. The idea in this study to evaluate the effect of above image features to query image. Again the comparison of Database image and query image is done on the basic of similarity measures such as Euclidian distance in above specified methods. Finally a threshold is obtained based on sample basis. We applied our approach to a database of 25 JPG images which includes 10 dinosaurs and 10 flowers. We have determined the capability of automatic indexing by analyzing image contents as features. The results are far more accurate than conventional image indexing. Hence, there exists a tradeoff between accuracy and computational cost.