Research On Deblurring Image In Low-light

With the development of science and technology,the quality of the image have been demanded increasingly.Especially in low-light condition,since a longer exposure time is often required in these cases to produce a well-lit image,the captured image using a handheld camera often ends up to be blurry due to inevitable camera shake.Then it seriously influences on people accessing to information and seeking the development of scientific research.Although significant progress has been made recently on image deblurring,some approaches often fail in low-light images.To study the new approach of deblurring in low-light images will be very important.Light streaks are common phenomena in low-light images that contain rich blur information.So this paper proposes a single-image,shift-invariant deblurring approach where the blur kernel is estimated from light streaks in the blurred image.Firstly,the edge of blurred image is enhanced by shock filter.And to perfom Laplacian of Gaussion(LOG)filtering thresholds on a blurred image to filter out isolated and sort edges.Which can reduce the dependence on recursion optimization algorithms.Secondly,defining a power-spectrum and the smallest distance to select light streaks.We introduce a non-linear blur model that explicitly models light streaks and their underlying light sources,and poses them as constraints for estimating the blur kernel.Experimental results show that this model can efficiently and effectively estimate the blur kernel,and the effect is better than others.Then,to discuss the image restoration.This paper combines Richardson-Lucy iterative algorithm with EM algorithm to restore latent images.So as to achieve a better restoration effect by using a simpler algorithm,and it is often more effective at suppressing ringing artifacts.By designing experiments and simulation,the results show that the method proposed in this paper has a deblurring performance significantly.