基于高斯曲率和局部方差的去噪模型

Noise removal model based on Gaussian curvature and local variance

用PM(Perona and Malik)模型去除椒盐噪声,使低噪声强度下未受噪的平坦区域的像素值减小,但是不能在有效去噪的同时保护纹理细节,导致图像模糊。为此,用局部方差和高斯曲率代替梯度模值来描述图像局部纹理细节,并定义了噪声度量函数,随之引入扩散方程,得到新去噪模型。实验结果表明:新模型不仅能有效地除去椒盐噪声和解决PM模型的问题,而且信噪比和峰值信噪比均有显著提高。因此新模型优于PM模型。

PM （ Perona and Malik） model uesd to remove the pepper and salt noise can reduce the pixel values of fiat region where noise in low intensity, but can not protect texture details while denoising, so local variance and Gaussian curvature in place of gradient modulus were used to describe local texture details of image, and noise-measured function was defined and introduced to diffusion equation, and then a new model was proposed. Experimental results show that new model not only removes noise and solves the problems of PM model effectively, but also both SNR and PSNR increase significantly.