Connected components objects feature for CBIR

Abstract

With the popularity of the network and expansion of multimedia technology, the traditional techniques of information retrieval do not satisfy the requirements of users. Recently, the content based image retrieval and its techniques have become the hot topic to satisfy a great development. In this paper, a new method is proposed to solve the problem of regions of interest (ROI) based image retrieval. The ROI technique which is based on segmenting the image into fixed partitions is computationally costly. The proposed method is based on the connected components and interesting of objects to generate the histogram and statistical texture feature vectors. These resulted vectors are used to retrieve images from a large image database. The color and texture features of the connected components are computed from the histograms of the quantized HSV color space and Gray Level Co-occurrence Matrix (GLCM), respectively. The vectors matching process is based on the histogram intersection. It is obvious the experimental data clearly shows the efficiency of the proposed method in comparison to the traditional ROI technique in terms of computationally cost.