CBRM: A New Content-based Regional-Matching Image Retrieval Method

A.H. Pilevar and M.T. Pilevar (Iran)


Image databases; content-based retrieval; Image Retrieval; region matching; wavelet


In this paper, we propose the content-based regional matching image retrieval (CBRM) method for retrieving general purpose images from databases, a novel CBIR algorithm that is robust to scaling and translation of objects within an image. The CBRM method is a novel method in which each image is first decomposed into regions. A measure for the overall similarity between images is developed using a region-matching scheme that integrates properties of all the regions in the images. In order to extract regions for an image, sliding windows are considered. An efficient dynamic programming algorithm is used to compute their signatures. The proposed method is tested using several databases, including a database of about 10,000 general purpose images. The experimental results obtained with the CBRM method, as compared to those obtained with the WBIIS and WALRUS methods, are satisfactory and demonstrate that it performs significantly better and faster than the other two.

Important Links:

Go Back