Patch-Wise Blind Image Deblurring Via Michelson Channel Prior

Guoquan Wen,Zhaozhong Yang,Xiaogang Li,Chao Min,Zeyun Jiang

doi:10.1109/access.2019.2959800

Abstract

Motion blur exists in many computer vision tasks, including faces, texts, and low-illumination images etc. It has been proved that Dark Channel Prior (DCP) and Bright Channel Prior (BCP) can both help the image deblurring by enhancing the dark or bright channel pixels. However, the pixels between the dark channel pixels and the bright channel pixels are not taken into consideration, which limits the deblurring performance. A novel image channel is proposed in combination with dark channel and bright channel in this paper to consider the effects of the all types of pixels, namely, Michelson channel pixels. Secondly, as the image channels are built based on the series of image patches with different blur kernels, a new method is developed to estimate the blur kernel and can measure the similarity between neighbored kernels. Meanwhile, to perform accurate kernel estimation, the L 0 regularization is applied into the algorithm framework. In the process of image deblurring, by enhancing the Michelson channels and retaining the other channels of the image, we can capture sharper image detail and eliminate the ringing artifacts of the recovered images. Massive experimental results demonstrate that the proposed method is more robust and outperforms the existing art-of-the-state of unsupervised image deblurring methods on both synthesized and natural images.

Highlights

In computer vision, the problem regarding to the image restoration including super-resolution, blur and noise removal is a classical vision problem, which has obtained significant progress in recent years [1]–[6]
We have studied the theory and effect of the dark channel prior and bright channel prior in the image deblurring
Based on the above observation, a new blind image deblurring method is presented in this paper, which is aiming at capturing Michelson channel pixels when highlighting dark channel pixels and bright channel pixels

Summary

INTRODUCTION

The problem regarding to the image restoration including super-resolution, blur and noise removal is a classical vision problem, which has obtained significant progress in recent years [1]–[6]. In [19], Jinshan Pan et al observe that enforcing the sparsity of the dark channel can help blind deblurring on various scenarios, including face, text and low-illumination images in the process of kernel and latent image estimation. What can be confirmed is that the sufficient dark pixels and bright pixels of the image can significantly affect the deblurring performances, which is similar to massive methods of image dehazing, like the work [21], [22]. Based on the above observation, a new blind image deblurring method is presented in this paper, which is aiming at capturing Michelson channel pixels when highlighting dark channel pixels and bright channel pixels. The effectiveness of the proposed method is validated by comparing the intensities of Michelson channel pixels on the clear images and the corresponding blurred image and comparing the deblurring performance of the other classical methods. According to the metric value, we can adjust the influence of different image patches on the estimated latent image by the corresponding weight parameters, and obtain the optimal solution

RELATED WORK

PROPOSED APPROACH

EXPERIMENTS

CONCLUSION