基于图像分解的多核非线性扩散去噪方法

研究了一种基于图像分解的多核非线性扩散去噪方法，利用两个非线性扩散模型分别提取图像的主信号和细节信息。先建立一个基于边缘定向的非线性扩散模型，实现对图像的主信号的提取。然后利用P—M扩散方程提取残余图像中的高频信号。将两步处理得到的信号进行合成，得到最后的处理结果。该方法能充分利用各个不同模型的优势，在整幅图像上均具有较好的处理效果。仿真计算结果表明，经该方法处理后的图像与现有的非线性扩散去噪方法相比，其噪声抑制更充分、边缘更清晰、峰值信噪比更高。

Multi-kernel nonlinear diffusion model for denoising based on image decomposition

A multi-kernel nonlinear diffusion model for image denoising based on image decomposition was studied, which got the major signal and the minutiae through different models. First,a edge-directed nonlinear diffusion model was put forward, which was used to get the major signal of image. Then P-M diffusion equation was used to get the high frequency signal of image from the remnant image. At last, the major signal and the minutiae were added to get the denoised image. The method can exert the merits of different models to get good results on the whole image. Simulation results show that, compared with the known methods, the model can remove noise more efficiently, keep edges more clearly, and it also has higher peak signal to noise ratio.