Blind deconvolution-based color image deblurring method

Abstract

The invention discloses a blind deconvolution-based color image deblurring method. The method is characterized by comprising the following steps of reading an original color blurred image I; extracting a MASK-I; performing refinement on a MASK-B by adopting maximal suppression; performing block segmentation on the MASK-B and the original color blurred image I; arranging pixels in each block according to a column vector form; performing summation on pixel values of column vectors; selecting effective column vectors for storage; clustering the column vectors; converting new column vectors into image blocks; converting new image blocks into a quasi clear image X1; extracting a MASK-X1 of the image X1; performing Fourier deconvolution to obtain a blurring kernel k1; and obtaining clear imagesof RGB channels, and performing iterative operation for 4 times to obtain a final color clear image. According to the method, the image deblurring effect is good; the deblurring speed is high; and themethod is simple and convenient to implement.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a method for deblurring color images based on blind deconvolution. Background technique [0002] Image deblurring is to take the existing blurred image obtained by people, and use a specific method to remove the blur in the image, so as to obtain the ideal clear image. In recent years, as people have higher and higher requirements for image clarity, deblurring methods are emerging, and the application fields involved are becoming more and more extensive. It will blur the image, so it is of great significance to deblur the image in daily production and life. To achieve a better processing effect, blurring caused by different reasons often requires different processing methods. Divided according to the blurred area, it can be divided into: global blur and local blur. The global blur is mainly caused by the motion of the shooting device, while the local blur is caused by the motion...

AI Technical Summary

Problems solved by technology

(2) The reason for the global blur is that the hand-held shooting device shoots with a relatively long exposure time, and often encounters blurring during the shooting process.

The L-R algorithm needs to consider the prior information of the object and does not need the noise information of the blurred image. It is suitable for linear and nonlinear imaging models. The solution is unique, but it needs to increase the number of iterations and limit the deviation threshold to achieve better results. Therefore, there is a certain limitations; the restoration effect of the constrained least squares algorithm is better than that of the Wiener filtering algorithm, but the derivation assumption of the constrained least squares algorithm is that the random field is uniform and the spectral density is known, so it is not suitable for wide application; blind The deconvolution algorithm can restore the image and the point spread function (fuzzy kernel) at the same time, and can still restore the blurred image without prior knowledge of the distortion, so it has good practicability, but which also produces a ringing effect

Images