Image deconvolution using homomorphic technique

Blind image deconvolution is a challenging issue in image processing. A solution to this problem is increasingly required in many applications. In this study, we develop a novel computational approach for solving the blind deconvolution problem by integrating the utilities of homomorphic domain and outlier handling methods in blurred images. Most of the existing methods for blind image deconvolution employ complex algorithms, and thus can incur excessive overhead in computing the blur kernel. In contrast, our work decomposes the blurred image into two main components using the homomorphic step. It is known that the homomorphic domain can be imposed on images by the logarithm operation that separates the image into the illumination and reflectance parts. The reflectance part contains the most prominent details of the image, while the illumination part contains mostly redundant information of the image. By using the reflectance part in the proposed blind deconvolution approach, we were able to achieve significant improvement in performance. The proposed approach outperforms the state-of-the-art methods. It is, therefore, an effective approach for blind image deconvolution with low complexity.