Adaptive kernel regression-based total variation image noise cancellation method

Abstract

The invention discloses an image noise cancellation method based on adaptive kernel regression and bilateral total variation. The invention mainly aims to solve the problem of failure to effectively maintain image detail information under a strong noise environment. The implementation process of the method includes the following steps that: 1, a contaminated image is obtained, a denoised image ofthe first iteration is initialized with the contaminated image; 2, the maximum number N of iterations is set; 3, the kernel regression weight value of an image of the t-th iteration is calculated; 4,the regular kernel of the image of the t-th iteration, which integrates with adaptive kernel regression and bilateral total variation, is constructed, and an energy functional is formed; 5, the derivative of the regular term to the iteration image is calculated; 6, an energy functional minimization problem is solved through adopting a steepest descent method, so that a denoised image of the (t+1)-th iteration is obtained; and 7, the step 3 to step 6 are repeated until the number of iterations reaches N, and a denoised image is outputted. With the method of the invention adopted, better targetimage texture, richer detail features and better visual effects can be maintained. The method can be used for the preprocessing of digital images.

Description

technical field [0001] The invention belongs to the field of digital image processing, in particular to a method for eliminating noise of a full variation image, which can be used for preprocessing of digital images. Background technique [0002] In the process of acquisition and transmission of digital images, due to the interference of the circuit itself and external noise sources, the quality will inevitably be degraded, which will seriously affect the subsequent feature extraction and analysis. Image denoising needs to take into account both the suppression of noise and the preservation of original image information. Aiming at the problem of image denoising, researchers in this field have carried out a lot of exploration and research, and proposed a large number of noise suppression methods based on digital signal processing technology. [0003] Existing image noise suppression methods mainly include: spatial domain noise suppression method and transform domain noise su...

AI Technical Summary

Problems solved by technology

However, in this method, when the size of the pixel-centered block is larger, the edges of the obtained image are less obvious, resulting in blurred denoised images.

Images