Half quadratic splitting method combined with convolution neural network for blind image deblurring
- PDF / 1,670,048 Bytes
- 16 Pages / 439.37 x 666.142 pts Page_size
- 30 Downloads / 221 Views
Half quadratic splitting method combined with convolution neural network for blind image deblurring Jiaqi Bao 1 & Lin Luo 1 & Yu Zhang 1 & Kai Yang 1 & Chaoyong Peng 1 & Jianping Peng 1 & Ran Li 1 Received: 26 February 2020 / Revised: 7 August 2020 / Accepted: 2 September 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
Blind image deblurring is the process of recovering the original image from a degraded image under unknown point spread function, and it is the solution to an ill-posed inverse problem. In this paper, the blurry image is firstly divided into skeleton image and blur kernel, aiming to achieve accurate blur kernel estimation. Then the advantages of modelbased optimization method and discriminative learning method are integrated through variable splitting technique. Finally, a trained convolutional neural network (CNN) is used as a module to be inserted into a model-based optimization method to solve the problem of blind image deblurring more effectively. By comparing visual and quantitative experimental data, the network proposed in this paper can provide powerful prior information for blind image deblurring and the restoration effects can approximate or exceed those of some representative algorithms. Keywords Blind image deblurring . Variable splitting technique . Convolutional neural network
1 Introduction Human beings rely on the visual system to obtain large amount of information. The research shows that about 70% of the information is obtained through the visual system, so it is particularly important to acquire, process and use the image information. Many image processing methods use image restoration technology to achieve the desired image effect. The significance of image restoration technology can be perceived from the space exploration more than 60 years ago. At that time, the images transmitted back to the earth from space were degraded due to unmatured imaging technology, unsatisfactory shooting environment, relative * Yu Zhang [email protected]
1
School of Physical Science and Technology, Southwest Jiaotong University, Chengdu 610031, China
Multimedia Tools and Applications
motion between objects and the jitter of the camera itself, which resulted in low image resolution and blurry image. To solve the problem of image degradation caused by various reasons, people began to study image restoration. During the production process and in real life, the two most typical image degradation phenomena are noise and blur. In the process of obtaining images, many factors will affect image quality, such as object motion, solar radiation, defocusing, optical deviation, and atmospheric [17, 23, 33, 34]. During the process of image transmission, the image will also produce blur and noise due to the interference of the transmission channel, the shooting of electronic components and other reasons. These degraded images bring great difficulties to subsequent image processing, including image feature extraction, target object tracking, etc. With the wide application
Data Loading...
