一种迭代滤波快速图像修复算法

为了充分利用图像局部区域信息来加快信息有效扩散及克服现有算法对图像纹理细节保护不足,提出了一种迭代滤波快速图像修复改进算法。该算法从破损区域由外向内取得一组有先后次序的待修复点,通过自适应选择其高斯掩膜领域,充分汲取已知（完好）区域信息,利用领域点等值线方向信息及远近距离来调整高斯核权系数,以准确控制滤波过程中的平滑强度;实验表明,与同类算法相比,该算法在同等迭代修复次数下,图像纹理更加清晰,峰值信噪比提高,具有较好的图像修复效果。

Algorithm for iterative filtering fast image inpainting

In order to make full use of the local area information of image to speed up information diffusion effectively and overcome the inefficiency of existing algorithms in the protection of image texture details when repairing,a novel iterative filtering fast image inpainting algorithm is proposed in this paper.Firstly,the proposed algorithm gets a group of points to be repaired in an order from outside to inside of damaged area,and the Gaussian mask exposure areas（field areas） of points to be repaired are adaptively selected according to their positions.Then the smooth strength of filtering process is accurately controlled by adjusting the Gaussian kernel weight coefficients with the directions of isophote of field points and the distances of them relative to the point to be repaired.The experimental results show that this algorithm can get a better subjective visual effect than the similar algorithms with the same iteration repairing times,and the repaired image is clearer and has a higher peak value signal-to-noise ratio.