Partial structure self-adapted image diffusing and de-noising method

Abstract

The invention provides a self-adaptive image diffusion denoising method of a local structure, which particularly relates to a classification method of an image local structure and a method for determining a diffusion conductivity coefficient according to the local structure. Image diffusion simulates the thermal diffusion form of the physics to carry out diffusion smoothing treatment to the pixelvalue of an image and realize the purpose of denoising. Anisotropic diffusion can retain detailed information, such as edge, line, and the like, when image smoothing is carried out. The anisotropic diffusion method that can regulate the diffusion strengths of different directions according to the local tiny structure of the image is provided by the invention. At first, pixel classification is carried out according to the local structure; different conductivity coefficient functions are defined for different types; and diffusion flow is figured out according to the functions, thereby realizingimage diffusion denoising under the condition of retaining edge information.

Description

technical field [0001] The invention belongs to the technical field of digital image processing in information technology, and relates to the extraction of classification feature values ​​of image pixels, a classification method of image pixels, and a calculation method of image diffusion flow. Background technique [0002] Image diffusion refers to the form of thermal diffusion that simulates physics, and performs diffusion and smoothing processing on image pixel values ​​to achieve the purpose of removing noise. [0003] The conductivity function refers to the specific function form used to calculate the diffusion flow according to the image gradient, and generally uses a non-monotonic negative decreasing function such as a negative exponential function. [0004] Gradient refers to the gray value difference between adjacent pixels in the image, which is a vector with a certain direction, including two components in the horizontal direction and vertical direction. [0005]...

AI Technical Summary

Problems solved by technology

However, the inconsistency defined in the algorithm cannot distinguish image structure information and noise very well, and it cannot effectively remove large-scale noise.

[0012] In the adaptive conduction coefficient function, the existing algorithms only consider statistics such as local gradient direction, variance, mean value, etc., without further and specific analysis of the local structure

Images