Blind Restoration Method of Turbulent Flow Image Based on Dark Channel Color and Alternating Direction Multiplier Method Optimization

Abstract

The present invention relates to a blind restoration method of turbulent images based on optimization of dark primary color and alternating direction multiplier method. First, based on the idea of ​​multi-scale, dark primary color prior constraints are applied to the image at each scale, and point spread function is imposed sparse constraints and energy constraints, and then the coordinate descent method is used to iteratively estimate the blur kernel and image at the current scale alternately. When the maximum scale is reached, the final estimated blur kernel is obtained. Finally, combined with the total variation model, the derivative alternating direction multiplier method is used to achieve rapid recovery of image details. The method of the present invention uses the dark primary color prior information of the clear image as a constraint item, which is beneficial to the cost function converging to a clear solution during the iterative process, and solves the problem that the blind restoration algorithm is easy to obtain using gradient prior information constraints under the maximum a posteriori probability framework. It solves the problem of fuzzy solution, so in the visual aspect of the restoration result, more image details can be restored and the ringing effect is less, which effectively improves the restoration quality.

Description

technical field [0001] The invention belongs to a digital image processing method, and relates to a new method for restoring single-frame atmospheric turbulent degraded images, in particular to a blind restoration method for turbulent images based on optimization of dark primary colors and alternating direction multiplier methods, which removes dark primary colors from fog in images The theory is applied to the field of turbulent image blind restoration, and the invention can be used in various military or civilian image deblurring processing systems. Background technique [0002] When the aircraft flies at supersonic speed in the atmosphere, it interacts violently with the atmosphere to form a complex high-temperature turbulent field. This turbulent effect will cause the target image received by the optical system of the aircraft to shift, shake, blur, etc., thereby seriously affecting Its ability to detect, identify and track targets, even fails to detect and identify targ...

AI Technical Summary

Problems solved by technology

Pan applied the dark channel prior theory in image dehazing for the first time to image deblurring, and achieved good results in processing motion blurred images, low-light blurred images and non-uniform blurred images, but the algorithm itself is sensitive to noise , when there is large noise in the blurred image, the processing result of the algorithm has a ringing effect; another way of thinking is to introduce edge selection and use strong edges to restore the image, but this kind of method involves complicated edge selection, how to design "big Gradient preservation and small gradient discarding" rule is a problem, and when the salience of the image is not very strong, the algorithm cannot select a suitable edge to estimate the blur kernel

[0005] It can be seen that the traditional blind restoration method of atmospheric turbulence image, in the case of serious noise or blurring, the visual quality of image restoration is poor, artifacts are prone to occur, and it is sensitive to noise.

Images