EXPLORE PUBLICATIONS BY COUNTRIES

CBIR, Row Mean, DCT, DST, Walsh, Haar, Heartley,, Slant, Kekre

The extension to the content based image retrieval (CBIR) technique based on row mean of transformed columns of image is presented here. As compared to earlier contemplation three image transforms, now the performance appraise of proposed CBIR technique is done using seven different image transforms like Discrete Cosine Transform (DCT), Discrete Sine Transform (DST), Hartley Transform, Haar Transform, Kekre Transform, Walsh Transform and Slant Transform. The generic image database with 1000 images spread across 11 categories is used to test the performance of proposed CBIR techniques. For each transform 55 queries (5 per category) were fired on the image database. Every technique is tested on both the color and grey version of image database. To compare the performance of image retrieval technique across transforms average precision and recall are computed of all queries. The results have shown the performance improvement (higher precision and recall values) with proposed methods compared to all pixel data of image at reduced computations resulting in faster retrieval in both gray as well as color versions of image database. Even the variation of considering DC component of transformed columns as part of feature vector and excluding it are also tested and it is found that presence of DC component in feature vector improvises the results in image retrieval. The ranking of transforms for performance in proposed gray CBIR techniques with DC component consideration can be given as DST, Haar, Hartley, DCT, Walsh, Slant and Kekre. In color variants of proposed techniques with DC component, the performance ranking of image transforms starting from best can be listed as DCT, Haar, Walsh, Slant, DST, Hartley and Kekre transform.