Damaged Image Restoration Based on Improved Criminisi Algorithm
Qi Guo, Jinhui Li · 2019
Criminisi algorithm has been widely used to repair damaged images, but the algorithm still has problems such as unreasonable priority selection algorithm and high error rate of fixed sample block matching. In view of these existing problems, based on the Crminisi algorithm, this paper adds edge terms, improves the priority algorithm, and sorts the image filling order reasonably. Experiments show that the modified Criminisi algorithm can overcome the shortcomings of the image discontinuity after repair, and can improve the quality of broken image restoration.