摘要
四阶各项异性扩散模型(AFOD)的提出有效解决了各向同性扩散模型(Y-K)的斑点效应、收敛速度慢等和边缘信息丢失等问题。但图像梯度作为边缘指示器,区分斜坡区域和边缘区域的能力弱,会使得去噪图像在斜坡区域产生阶梯效应。本文利用差分曲率能有效区分图像边缘区域和斜坡区域的特点,将AFOD模型和P-Laplace结合,使得模型能在保护图像边缘的同时,有效抑制阶梯效应。实验结果与几种相关方法进行比较,证明了该模型稳定良好的性能。
The Anisotropic Fourth-Order Diffusion Filter (AFOD) is proposed to effectively solve the problems of speckle effect, slow convergence speed and loss of edge information of isotropic Diffusion Filter (Y-K). However, as an edge indicator, the image gradient has a weak ability to distinguish the slope area from the edge area, which will cause the denoised image to produce a staircase effect in the slope area. In this paper, the difference curvature can effectively distinguish the edge area and the slope area of the image, and the AFOD model is combined with P-Laplace, so that the model can effectively suppress the step effect while protecting the image edge. The experimental results are compared with several related methods, which demonstrate the stable and good performance of the model.
出处
《理论数学》
2022年第11期1981-1988,共8页
Pure Mathematics
