Scalable region-based image retrieval system in the Wavelet Transform domain

In this paper, we have designed a new method for region based retrieval of textured mono-channel images. To perform segmentation and region feature extraction, the concept of Markov Random Fields (MRF) in the Wavelet Transform (WT) domain is considered and the Iterated Conditional Modes (ICM) algorithm is employed. The novelty of our approach consists in exploiting the hierarchical organization of features through scales and iterations in order to design a multi-resolution and iteratively refined search procedure until user satisfaction. Experimental results show that our method achieves a suitable tradeoff between complexity and reliability of the search procedure.