Turbulence-degraded image blind restoration method based on edge prediction and sparse ratio regular constraints

Abstract

The invention relates to a turbulence-degraded image blind restoration method based on edge prediction and sparse ratio regular constraints. The method is technically characterized by comprising the steps of predicating an effective edge in a current image to be restored, combining edge predication information with sparse prior information of a natural image edge to guide restoration of a point spread function, restoring the current target image according to a non-blind restoration algorithm, regarding the restored image as input of edge predication of the next time, and carrying out the iterative cycle in this way till a clear restored image is obtained. According to the method, by combining the prior information of an image with effective information contained in the degraded image, artifacts generated in the image restoration process can be effectively restrained, more details can be restored, and the restoration effect is better.

Description

technical field [0001] The invention relates to a blind restoration method of turbulence degraded image based on edge prediction and sparse ratio regular constraint, that is, blind restoration of turbulence degraded image based on edge prediction and sparse regular term constraint. Applying sparse representation theory to the field of blind restoration of turbulent images, the invention can be applied to various military or civilian image processing systems. Background technique [0002] Atmospheric turbulence is the main cause of image degradation in astronomical observations. Changes in the refractive index of air caused by turbulence can lead to random fluctuations in the amplitude and phase of light waves, resulting in phenomena such as light intensity scintillation, wave front distortion, and beam drift. The essence of the influence of turbulence on light wave propagation is to change the original wave front of the light wave. The wave front is distorted from a plane to...

AI Technical Summary

Problems solved by technology

In 2011, Krishnan research pointed out that l 1 / l 2 Norm ratio l 1 The norm is closer to l 0 Norm, with l 1 / l 2 As a constraint, a more realistic solution can be obtained, but Krishnan uses l 1 / l 2 The solution sought as a constraint has severe artifacts

And most of the methods based on sparse regularization constraints directly restore the gradient image of the degraded image as the edge-guided point spread function of the natural image. Due to the serious blur and noise in the degraded image, the gradient image obtained must contain many false edges. These false edges will mislead the point spread function restoration results, resulting in more artifacts in the restored image

[0003] In the case of degraded images with serious noise or blur, the previous blind image restoration algorithms tend to cause serious artifacts in the restored image, making it difficult to obtain satisfactory results for the restored image.

Images