Double Cross Algorithm of Improved TV Image Inpainting Model

Currently, the numerical solution of most variation inpainting models only uses four neighborhood information of damaged pixel. Due to the less reference information provided, the inpainting accuracy is low. This paper presents double cross algorithm based on improved Total Variation (TV) model. The algorithm uses eight neighborhood information of damaged pixel. Firstly, it divides the eight neighborhood pixels into two groups. Secondly, each group uses four neighborhood pixels to compute the pixel value of damaged pixel. Finally, the weighted mean of these two pixel values is the final inpainting pixel value. The experiments show that the new algorithm significantly improves the inpainting accuracy.