An Efficient Approach for Region-Based Image Classification and Retrieval

In tins paper, a fast and efficient approach for region-based image classification and retrieval using multi-level neural network model is proposed. The advantages of this particular model in image classification and retrieval domain will be highlighted. The proposed approach accomplishes its goal in two main steps. First, by aid of a mean-shift based segmentation algorithm, significant regions of the image are isolated. Then, features of these regions are extracted and then classified by the multi-level model into five categories, i.e., "Sky", "Building", "Sand\Rock", "Grass" and "Water". Features extraction is done by using color moments and 2D wavelets decomposition technique. Experimental results show that the proposed approach can achieve precision of better than 93% that justifies the viability of the proposed approach compared with other state-of-the-art classification and retrieval approaches.