一种结合灰度和梯度方向的图像修复方法

Image inpainting by combining gray levels and gradient directions

描述了一种结合灰度和梯度方向的图像修复方法。通过优化一个图像灰度和梯度方向的能量函数来实现修复。在定义能量函数时,综合考虑了灰度、梯度和曲率,并用梯度方向角来表示曲率,这降低了要求解的偏微分方程阶次。方法具有较好的图像平滑能力,符合人眼的连通性原则,并能较好地修复物体的拐角。

This paper proposes an image inpainting scheme by combining gray levels and gradient directions. The inpainting processing performs with optimizing an energy function of image gray and gradient directions. In the energy function, three factors including gray, gradient and curvature are considered simultaneously. The curvature is expressed with the angle of gradient direction, which reduces the order of partial differential equations to be solved. The scheme has strong capability of image smoothing. It can meet the connectivity principle of human vision perception and restore object comers satisfactorily.