Image Inpainting Based on Improved Criminisi Algorithm

To solve the problem of unreasonable priority repairing order of traditional Criminisi algorithm, this paper proposes an image repairing algorithm with improved priority Criminisi. Firstly, the information entropy, which measures the complexity of the image structure, is introduced into the calculation of the priority to integrate the structural information of the known part of the block to be repaired, to enhance the influence of the structural information of the image on the priority. Then, the priority function is redefined as the product of the confidence term plus data term and information entropy, which makes the restoration order of the image more reasonable. The experimental results show that the improved Criminisi algorithm can improve the repair quality of images with different complexity.