论文部分内容阅读
图像的非线性扩散滤波基于热扩散方程,其关键是确定适当的扩散量与扩散方向.在已有的滤波模型中,扩散系数依赖于梯度,因而容易受到噪声的干扰.为了克服这一缺陷,首先给出了一种新的计算扩散系数的多尺度方法.然后建立了一类用于图像恢复的非线性小波扩散(NWD)模型.NWD模型有较强的抗噪能力,但同已有的模型一样,需要较大的计算量.为此,通过NWD模型的简化,给出了线性小波扩散(LWD)模型.LWD模型由输运与扩散两部分组成.由于存在输运,所以其滤波是各向异性的,能够较好地保持边缘,尽管在边缘点处扩散是各向同性的.LWD模型的优点是简单易于分析,具有较小的计算量.最后给出了所提模型与已有模型相比较的数值实验结果.
Based on the thermal diffusion equation, the key to the nonlinear diffusion filtering of the image is to determine the appropriate diffusion and diffusion direction.In the existing filtering model, the diffusion coefficient depends on the gradient and is therefore susceptible to noise.In order to overcome this shortcomings, Firstly, a new multi-scale method for calculating diffusion coefficient is given, and then a class of nonlinear Wavelet Diffusion (NWD) model for image restoration is established.NWD model has strong anti-noise ability, but with the existing Which needs a large amount of computation.For this reason, a linear wavelet propagation (LWD) model is given by simplifying the NWD model.The LWD model consists of two parts: transport and diffusion.Due to the existence of transport, the filtering is Anisotropy, which can keep the edge well, although the diffusion is isotropic at the edge point.The advantages of the LWD model are simple and easy to analyze, with a small amount of computation.Finally, the relationship between the proposed model and the existing Numerical comparison of the experimental results.