全变分自适应图像去噪模型

通过分析三种主要变分去噪模型(调和、全变分以及广义全变分模型)的优缺点,提出了一种基于全变分的自适应图像去噪模型。该模型根据噪声图像的信噪比,采用高斯滤波器对图像进行预处理,克服了全变分模型引入的阶梯效应;利用图像中每一像素点的梯度信息,自适应选取去噪模型中决定扩散强弱的参数p(x,y),使接近边缘处平滑较弱,远离边缘处平滑较强。数值实验表明,本方法在去除噪声的同时保留了图像的细节信息,取得了很好的降噪性能,其峰值信噪比(PSNR)在高噪声水平下,较其他变分方法至少提高1.0dB左右。

Adaptive image denoising model based on total variation

An adaptive image denoising model based on Total Variation (TV) is proposed by analyzing the three important denoising models: harmonical model, TV model and generalized TV model, in the variational image restoration. Firstly, the convolution of the Gaussian filter and the noisy image can remove a small portion of the noise so it is less likely to be detected as an edge, and then we can adaptively select the most appropriate denoising scheme based on the gradient information of each pixel. Numerical experiments show that the proposed method can remove the noise while preserving significant image details. At high noise level,the method achieves at least 1.0dB gain over other variational denoising methods for Peak Signal-Noise Ratio (PSNR) measurement.