改进曲率驱动模型的敦煌壁画修复算法

针对敦煌壁画裂纹形状复杂、划痕不规则,采用CDD曲率扩散算法修复时易出现假边缘、阶梯效应以及修复时间长的问题,提出一种改进曲率驱动模型的自适应敦煌壁画修复算法.首先对CDD算法中梯度消失问题进行改进,使扩散项更加合理;然后引入自适应控制策略,壁画破损区域曲率不同时能够选择不同的修复模型进行动态自适应控制修复;最后利用光滑函数进行角点扩散修复,达到去噪和保护边缘的目的.对真实敦煌壁画进行数字化修复的实验结果表明,该算法较好地解决了CDD算法的边缘过渡不自然以及修复时间过长的问题,在PSNR和修复时间评价指标上均优于其他对比算法.

Inpainting Algorithm for Dunhuang Mural Based on Improved Curvature-Driven Diffusion Model

The cracks of Dunhuang mural have the characteristic of complex shape and irregular scratches,and it is prone to false edges,staircase effects and long repair time by using curvature-driven diffusion(CDD)algorithm.An adaptive inpainting algorithm for Dunhuang mural based on improved CDD model is proposed in this paper.Firstly,the gradient disappearance problem in CDD algorithm is improved to make the diffusion term more reasonable.Then an adaptive control strategy is proposed.Different repair models can be selected to dynamically self-adaptively repair when the curvature of the damaged area of mural crack is different.Finally,a corner diffusion repair is carried out by using smooth function to achieve the purposes of denoising and edge protection.The experimental results of restoration of real Dunhuang mural show that the algorithm can solve the problems of unnatural transition edges and long repair time of CDD algorithm.The proposed method is superior to other algorithms in PSNR and restoration time evaluation.