The Research Of Image Restoration Method Based On Improved Grey Wolf Algorithms-BP Neural Network

With the continuous development of information technology,multimedia technology has been applied to all areas of life.As the main carrier of digital multimedia content,visual signals such as images and videos play an irreplaceable role in the process of video communication and visual perception.However,due to the influence of factors such as motion,jitter,and electronic interference,the image will be degraded during shooting,transmission,and storage.Image restoration is the inverse of image degradation,which is to make the restored image as close as possible to the original image.Image restoration technology is one of the research hotspots in the field of digital image processing.It has important theoretical and practical significance for the reproduction of degraded images.This paper first introduces the degradation/recovery model,and then introduces the traditional image restoration techniques,such as Wiener filtering,inverse filtering,LR restoration,least squares filtering and restoration methods,and the restoration experiments of degraded images,but the effect of restoration It is not ideal.There are a lot of white points in the image restored by Wiener filtering and inverse filtering,and the restored image of LR restoration and least squares filtering is blurred.In order to make the restored image closer to the original image closer to the original image,this paper uses BP neural network to have the characteristics of learning and generalization,and BP neural network as a recovery tool for image restoration.This restoration method can be used in images.Get rid of the shortcomings of traditional restoration methods that rely too much on prior knowledge.Although the BP network algorithm can effectively recover the degraded image,the BP recovery network has the disadvantages of slow convergence and high initial parameters in the learning process.Aiming at the shortcomings of BP recovery network,this paper proposes to use the strong global search capability of Grey Wolf algorithm to eliminate its excessive dependence on initial parameters and improve network performance.In order to further enhance the optimization ability of the grey wolf algorithm,the convergence speed is improved by the convergence factor,the dynamic weight update method is used to enhance the environment adaptability,and the differential optimization algorithm is used to enhance the global search ability of the gray wolf algorithm.The restoration experiments of the degraded images verify that the improved algorithm has better evaluation criteria such as image structure similarity,peak signal-to-noise ratio and normalized mean square error than the BP restoration network and the classical restoration algorithm.