An Effective Image Retrieval System using Region and Contour based Features

In this paper, a hybrid approach is proposed for improving the image retrieval accuracy. In the hybrid approach both local and global features of images are combined, which represent the entire aspects of images. Local features are extracted using Fourier descriptors and global features are extracted by means of angular radial transform. The results of combining both these descriptors demonstrate that the proposed solution provides significant improvement in image retrieval accuracy as compared to using global and local features autonomously. Besides, it also outperforms existing recent hybrid approaches to image retrieval.