基于小波变换模的FAB扩散模型

Forward-and-backward Di?usion Processes Based on Wavelet Transform Modulus

将小波多尺度边缘检测的思想引入到前向-后向(FAB)扩散模型中,根据小波变换能够准确提取图像局部结构信息的特性,给出了利用多个尺度上小波变换模构造图像扩散系数的方法,进而建立了基于小波变换模的前向-后向扩散模型.最后给出了一系列数值实验结果,以比较本文方法与传统FAB方法的图像平滑增强效果.

This paper introduces the idea of the multiscale edge detection via wavelet modulus into the forward-and-backward（FAB） diffusion models.By utilizing the characteristics that the wavelet transform can well extract the local structure information of an image,we present a new method to construct an diffusivity using wavelet modulus at several scales,and establish the wavelet-based FAB diffusion model.Finally,we present several numerical experiments for image smoothing and enhancing to compare the proposed model with the previous FAB models.