基于网函数插值的图像修复算法

Image Inpainting Algorithm Based on Net Function Interpolation

图像在成像、传输等过程中会因为特定因素而导致图像数据失真,需对其进行修复。基于网函数插值理论,利用插值实现扩散修复,构造网函数插值图像修复算法,对斑点、划痕、文字三种不同类型的缺损图像进行修复实验,并与全变分(Total Variation,TV)修复算法进行对比实验。实验结果表明:1)该算法可有效修复三种不同类型的缺损图像;2)算法计算复杂度低,在多数情况下比TV修复算法收敛速度更快,修复效果更好。

In the process of imaging and transmission,the image data will be damaged due to some factors,and it needs to be repaired.The net function interpolation theory is used to achieve diffusion recovery,and the net function interpolation image restoration algorithm is constructed to perform experiments on three different types of defect images of spots,scratches and text.The total variation(TV)repair algorithm is used to compare with it.The experimental results show that the algorithm can effectively repair three types of damaged images.The algorithm has low computational complexity.In most cases,it’s convergence rate is faster than the TV repair algorithm,and the repair effect better.