分析了进行图像放大时各向异性偏微分方程优于各向同性偏微分方程,分析了不同四阶模型的扩散方向。为了消除低阶偏微分方程在处理图像中出现的块状效应的影响,同时保证方程为各向异性扩散,构造了两个各向异性的四阶偏微分方程,并且分别从数据和放大图像效果两方面来说明本文模型优于其他4个模型。
In this paper, we show that for image zooming, the anisotropic partial differential equation is better than isotropic partial differential equation. Then we analyze the diffusion directions of the different fourth-order PDEs in this paper. While ensuring that the equation is anisotropic diffusion and in order to eliminate the block effects of low order partial differential equations in image processing, we construct two anisotropic fourth-order partial differential equations. From the data and qual- ity of image zooming, we conclude that the models we present are better than the other four models.