一种新的基于扩散方程的图像平滑方法

针对退化扩散方程在对图像平滑时,奇异点容易被破坏,引入了小波变换模极大值表示,提出了由小波变换模M(s,x,y)决定扩散速度的图像平滑方法.该方法在图像奇异点的位置使平滑速度趋于零;在图像的光滑区域扩散速度由小波变换模决定;在灰度变化不大的区域,采用各向同性扩散.该方法克服了退化方程对图像奇异点的模糊现象,兼顾了去噪和保边界.实验证明该方法优于其他的方法.

A new image smoothing method based on diffusion equation

Due to the singular points of degenerate diffusion equation can easily be destroyed when image is smoothed, the representation of maximum value of wavelet transform module M(s,x,y) is developed. It is showed that the diffusion velocity is dominated by the wavelet transform module . Under this method, the diffusion velocity tends to zeros at the singular points, the diffusion velocity is dominated by the wavelet transform model at smoothing points. The diffusion is isotropic at points where the gradient is low. This method overcomes the fuzzy phenomenon when the degenerate diffusion equation smoothed the image at the singular points. Meanwhile, the noise is eliminated and the edge is well preserved. The experiment shows that this method is better than others.