Blind Motion Deblurring For Single Images

Image is an important medium for information transmission,widely used in our daily production and life.However,the relative motion between the imaging device and the target object is likely to occur during image capturing,causing the captured image to be blurred and seriously restricting the use of the image.The purpose of image motion deblurring is to use a numerical method to recover a clear image from a degraded image.Image motion deblurring is an important branch of image processing,especially in recent years,the rise of artificial intelligence.Its research has an extremely important scientific value and practical significance.By whether the blur kernel is known,we can divide the image motion deblurring into two categories:unblind and blind motion deblurring.The problem of image blind motion deblurring is difficult to solve because of its unknown quantity is much more than known quantity,which has serious ill-posed.But most of the blurred kernels of motion blurred images are unknown,so it is necessary to study the problem of image blind motion deblurring.This paper focuses on the blind motion deblurring problem of single images,and the main work includes the following aspects:(a)Based on the analysis of the statistical features of natural images,we propose to apply the image second-order gradient feature to image blind motion deblurring for the first time.On the basis of previous studies,we propose a new method of natural image blind motion deblurring based on image second-order gradient and normalized sparse prior.An effective fast algorithm is given.A series of experimental results show that this method can effectively suppress the staircase effect phenomenon in the restoring images of other existing algorithms while removing the natural motion blur;(b)Due to special images,such as text images,do not meet the statistical characteristics of general natural images,it is often necessary to design separate deblurring methods.On the basis of previous studies,we propose a new blind motion deblurring method based on image second-order gradient and dark channel sparse prior that is effective for both natural images and special images.In this thesis,we validate the effectiveness of our methods with a series of experiments and compare them with other existing methods.The experimental results show that the proposed methods can effectively remove the image motion blur and improve image quality.The works of this paper make a corresponding improvement to the problem of image motion deblurring between natural images and special images.