Scatter matrix feature-based second order partial differential equation remote sensing image denoising method

Abstract

The invention discloses a scatter matrix feature-based second order partial differential equation remote sensing image denoising method, and belongs to the field of image processing. Firstly, a control function of diffusion terms is established by introducing a 45N degree rotation module value invariance operator, so that diffusion is performed in a gradient direction and a vertical gradient direction in a smooth region and is performed in a curvature direction only in an edge region; and secondly, a better distinguishing capability for feature regions of an image is realized through a scattermatrix, and based on this, fidelity terms are created, so that the denoising effects of a sharp corner region and the smooth region are improved, and the over-smoothness effect of an MCM at a targetsharp point and the "blocky" effect generated by a PM model are effectively relieved.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a second-order partial differential equation remote sensing image denoising method based on the characteristics of a scatter matrix, which has strong ability to maintain edges and sharp corners and high diffusion efficiency in smooth areas. Background technique [0002] Remote sensing image is a special type of image obtained by using the radiation characteristics of the observed object and the reflectivity of electromagnetic waves. It has become an important means of earth observation, involving many fields of national politics, economy, military and society and playing an important role. . However, because satellite remote sensing systems are often in complex electromagnetic wave environments, or due to factors such as electromagnetic interference between load components or periodic offsets of sensors, remote sensing images will be affected by noise to varying degrees during the...

AI Technical Summary

Problems solved by technology

However, the smoothing efficiency of this type of method for strong edge noise and isolated noise such as salt and pepper noise is not satisfactory, and many scholars have found that the PM method is "sick" and cannot guarantee the existence and uniqueness of the equation solution

For this reason, Catte et al. regularized the PM model to solve the uniqueness of the solution of the equation, but it is difficult to ensure the stability when the variance tends to 0

In order to better preserve the edge characteristics of the image, Alvarez et al. proposed the MCM (Mean Curvature Model) diffusion model based on the average curvature, so that the diffusion is always along the direction perpendicular to the gradient, but due to the diffusion speed and curvature of the model Proportional to , it will produce excessive smoothing in the sharp corner area of ​​the image

Images