Single lens imaging method for extracting blurring kernel priori according to image spectrum information

Abstract

The invention relates to the field of image processing, and particularly discloses a single lens imaging method for extracting a blurring kernel priori according to image spectrum information. The method comprises the steps that a single lens camera is used for acquiring a blurred image; an l1 / l2 image priori and the blurring kernel priori extracted according to the information of the blurred image are added in an existing target optimization function; then a blind convolution algorithm based on maximum posterior probability is used for estimating the blurring kernel of the single lens camera; combined with the estimated blurring kernel, a non-blind convolution algorithm is used for obtaining a clear image. According to the single lens imaging method for extracting the blurring kernel priori according to the image spectrum information, single lens high-quality computational imaging can be achieved, the blurred image information can be fully used by the adopted blurring kernel, and therefore a good deblurring effect is achieved.

Description

Technical field: [0001] The invention belongs to the field of image processing, and mainly relates to image deblurring, in particular to a single-lens imaging method for extracting fuzzy kernel priors according to image spectrum information. Background technique: [0002] A single lens is the most basic component of a modern optical imaging system such as a SLR (Single Lens Reflex, SLR) lens. With the continuous improvement of people's requirements for image quality, the precision of the lens is also increasing. Increasingly complex lens designs are aimed at compensating for the aberrations (Aberrations) caused by single lenses, which are the deviations between the actual optical system and the ideal optical system, resulting in blurred images. Optical aberrations include chromatic aberration, spherical aberration, coma, astigmatism, curvature of field, and distortion [1]. All spherical single lenses will produce the above-mentioned optical aberrations, which cannot be dire...

AI Technical Summary

Problems solved by technology

[0028] In view of the above-mentioned problem that the existing blur kernel prior cannot meet the requirements of single-lens computational imaging, the purpose of the invention is to propose a single-lens imaging method for extracting the fuzzy kernel prior based on image spectrum information, combined with a new blur kernel prior , using the non-blind convolution image restoration algorithm to obtain a clear image after deblurring, in order to achieve better deblurring effect

Images