Energy functional based image smoothing and sharpening algorithm

Abstract

The invention relates to an energy functional based image smoothing and sharpening algorithm. The energy functional based image smoothing and sharpening algorithm comprises the following steps of 1 converting an anisotropic diffusion equation into a minimized energy functional, 2 representing a dynamic equation in an internal coordinate mode, 3 establishing the energy functional (shown in the description), 4 establishing a gradient threshold function k = e - at of an image, 5 introducing a fidelity item (shown in the description) to establish the energy functional based image smoothing and sharpening algorithm and 6 adopting a central difference numerical algorithm to further process a result obtained in the step 5. By means of the energy functional based image smoothing and sharpening algorithm, information amount required by the algorithm is low, the algorithm is simple, image denoising is achieved, and polluted images are more approximate to original images.

Description

Technical field [0001] The invention relates to the field of image denoising algorithms, in particular to an image denoising algorithm based on energy functionals. Background technique [0002] The main sources of image noise are random Gaussian noise during image acquisition and salt and pepper noise during image propagation. Traditional denoising methods include median filtering, homomorphic filtering, inverse filtering, etc. These methods can achieve the purpose of removing noise to a certain extent, but they have a common weakness. When denoising, they will also make the image The edges are blurred, and even the detailed texture information of the image is lost. Compared with traditional methods, image processing methods based on partial differential equations have stronger local adaptive capabilities and higher flexibility, and have important applications in image denoising, segmentation, edge detection, and enhancement. [0003] The image processing method based on partial ...

AI Technical Summary

Problems solved by technology

[0003] The image processing method based on the partial differential equation is based on the continuous mathematical model of the image, so that the image changes according to a specified partial differential equation. The most representative algorithm based on the variational model is the nonlinear partial differential equation denoising algorithm ( TV algorithm), which classifies the image into a variational bounded function space, uses the first-order norm full variation as its "smoothness" measure, and constrains along the direction perpendicular to the gradient, which well protects the image edge, but because it does not fully conform to the morphological principles of image processing, it usually produces a staircase effect after multiple iterations

Images