Home > Published Issues > 2019 > Volume 7, No. 1, March 2019 >

Improvement of Robustness in Blind Image Restoration Method Using Failing Detection Process

Takahiro Nagata1, Tomio Goto1, Satoshi Motohashi2, Haifeng Chen3, and Reo Aoki3
1. Dept. of Computer Science, Nagoya Institute of Technology, Nagoya, Japan
2. Tokai Rika Co. Ltd., Toyota, Japan
3. R&D, Visual Technologies (ASIC), EIZO Corporation, Ishikawa, Japan

Abstract—Blind image restoration, which restores a clear image from a single blurry image, is a difficult process of estimating two unknowns: a point-spread function (PSF) and an ideal image. In this paper, we propose a novel blind deconvolution method to alternately estimate a PSF and its latent image. We apply a gradient reliability map that enables edge selection appropriate for PSF estimation and an energy function that enables estimation of convergence states. This method improves restoration performance by eliminating noise adversely affecting estimation. Additionally, a restoration failure detection process is added by using an evaluation function. Experimental results show that the robustness of the proposed method is improved and high quality images are obtained.

Index Terms—blur, blind deconvolution, image restoration, point spread function

Cite: Takahiro Nagata, Tomio Goto, Satoshi Motohashi, Haifeng Chen, and Reo Aoki, "Improvement of Robustness in Blind Image Restoration Method Using Failing Detection Process," Journal of Image and Graphics, Vol. 7, No. 1, pp. 1-8, March 2019. doi: 10.18178/joig.7.1.1-8