通过使热传导的传播方向及强度与图像内容相关联,提出了一种基于各向异性热传导模型的图像修复算法。该算法的基本思想是:令等照度线方向的传播强度恒为常数,而沿梯度方向信息的传播强度与图像梯度的模成反比,这样在实现结构信息修复的同时,有效解决了边缘区域模糊的问题。在数值计算中根据曲率的大小来定义结构的复杂程度,自适应地选择半点或整点差分格式,提高了处理效率。相比于其他高阶偏微分方程修复技术,该方法的数学表达更加简洁,且修复结果可取得较高的峰值信噪比以及视觉上满意的效果。
We propose an efficient image inpainting method using anisotropic heat transfer model.In the method,the propagating direction and intensity are related to the image contents.The propagation strength along gradient direction is made inversely proportional to the magnitude of the gradient,which avoids the blurring effect near edge area.In numerical implementation,the step size of finite difference is adaptively chosen according to the curvature,which leads to satisfactory inpainting quality and fewer iteration steps.Compared with other high order PDE methods,the proposed method is more concise,and can give better results for structure repairing.Experiments were carried out to show the effectiveness of the method.