基于全变差和P-Laplace模型的混合图像修复算法

图像修复是近年来图像处理研究的主要问题之一.在基于偏微分方程的修复算法中,全变差(total variation,TV)模型能够很好地保护图像边缘信息,但其各向异性扩散方式在平坦区域容易产生阶梯效应;而在图像平坦区域具有良好修复效果的P-Laplace模型,其各向同性扩散方式不适于修复图像边缘信息.将TV模型和P-Laplace模型有机结合起来,提出了一种混合图像修复算法.提出的扩散控制参数k能够根据待修复像素所在区域调节两种信息扩散方式的重要程度,实现混合图像修复.实验结果表明,所提算法获得了更好的修复结果.

Hybrid image restoration algorithm based on total variation and P-Laplace models

Image restoration is one of the major problems of image processing research in recent years. In the image restoration algorithms based on partial differential equation, the total variation (TV) model can well protect the image edge information, but in the flat areas, the anisotropic diffusion TV model can easily generate ladder effect. While the isotropic diffusion P-Laplace model can obtain good restoration results in the flat areas, but it is not suitable to restore the image edge information. Based on the TV and P-Laplace models, a hybrid image restoration algorithm is proposed, in which the control parameter k can adjust the importance degree of the two diffusion methods according to the image areas, and realize hybrid image restoration. Experimental results show that the proposed algorithm can obtain better restoration results.