A RWSM based Efficient CBIR system in JPEG Domain with GMCO Algorithm
Abstract
Content-Based Image Retrieval (CBIR) is an efficient system for searching the desired image from a large database with the help of the various statistical features extracted from the image. The existing methods developed for image retrieval have failed to meet the user requirements, mainly due to the inability of retrieving accurate images and consumes more time for retrieving. To overcome these issues, this paper proposes an RWSM similarity calculation measure and global and local features for retrieving the images. Firstly, the JPEG images are used as the input images and are decomposed by using the RDTCWT algorithm. After that, color and texture features are extracted from the images by using the coefficient of sub bands of RDTCWT. The same processes are done for the query image too. For retrieving very similar images, the similarity is calculated in between the obtained class of database image and the query image features by using the RWSM method, and then, the higher score image is retrieved as the output. At last, the user is satisfied with the images that are retrieved using the GMCO method as relevance feedback. The experimental evaluation of the proposed method demonstrates that the proposed image retrieval system achieves better results than the existing methods.