一种新的变分消噪理论和算法研究

文章从梯度下降法解相应的欧拉方程着手,提出一种新的定参图像去噪方法.新方法在选取参数的同时又保证均值和方差估计式,并且得到了解偏微分方程新的初值.这个初值是噪声图像经过小波分解后的原低频系数和设置阈值的高频系数的重构图像.数值仿真结果表明,该方法选取的初值具有更好地去噪效果.

Resaerch on the New Theories and Algorithms of Variation Denoising

This paper proposed a newly-selected parameter method when solving the corresponding Euler e-quation by the gradient method in denoising image,and realized the mean and the variance estimate formula simultaneously by adopting this method. The initial value was the reconstructed image of the noise image, under the wavelet decomposition,keeping the low frequency coefficients and only setting up a threshold for high frequency coefficients. With the newly-selected initial value,this paper solved PDE by the finite difference iteration and stimulated the equation numerically. The experiment results showed that this method had a better denoising effect.