A Novel Technique for Region-Based Features Similarity for Content-Based Image Retrieval

The study of feature extraction and clustering techniques for an image is carried out to find the future scope in the area of image clustering for the real world applications. Nowadays, a huge amount of image data is being collected in various application domains. As clustering approaches group homogeneous data together and also deals with unlabeled data, it is used quite often for accessing the interesting data easily and quickly. Image clustering is a process of grouping image data on the basis of similarities present between them. Features extracted from the images are used for the computation of similarities among them. This paper presents a technique for CBIR (Content Based Image Retrieval) by selecting the regions on the basis of their contribution to image contents. Texture and edge features are extracted at region-level whereas shape feature is extracted at image-level. At region-level image is divided in nonoverlapping regions. Texture and edge features are calculated for each region separately. Curvelet transform is used for extracting the texture feature by providing the curve continuity as well as line continuity in the feature extraction process. Moment invariant is used for extracting the shape features. All of the regions of the image may not have equal contribution in identifying the user perception of the image. Proposed method does not dominate to the non-highlighted regions but it decreases the region weight for less contributing regions. IRM (Integrated Region Matching) technique is used for retrieving the relevant images. The performance of the proposed system is tested on Flickr and COREL databases. Experimental results show that the retrieval performance of the proposed algorithm is better in comparison to other state-of-the-art methods.