基于非线性扩散滤波结构信息的图像去噪方法

Image denoising method based on structure information of nonlinear diffusion filtering

为在去除噪声的同时保持更多的边缘信息,利用拉普拉斯算子的旋转不变性提出一种适用于更多角度的拉普拉斯算子模板,根据模板适应角度的不同提出相应的权重函数,使非线性扩散滤波在原有的去噪基础上能够自适应地选取对应的算子对含噪图像进行扩散处理。实验结果表明,该方法在去除噪声的同时可以保持更多的边缘信息。

To retain more marginal information while denoising,by utilizing the rotational invariance of Laplacian kernel,a Lapacian kernel was proposed to fit different angles and the corresponding weighting function was put forward to make the nonlinear differential filter self-adapted.Experimental results show that the proposed method is able to enhance the edge-preservation capability.