In this work we propose a new inpainting algorithm for color images. It is a patch-based algorithm that replicates some small areas across the image into the unknown area, in order to obtain a complete image with no visual differences between the original part and the reconstructed one. In this proposal we use color and gradient features to calculate the similarity between small windows of the image. The results show that our algorithm is able to obtain final results with better textures than the ones that only take into account the color features.

Our new inpainting algorithm for color images, using color and gradient features, produces better textures and results compared to those only considering color features.

2017 Joint 17th World Congress of International Fuzzy Systems Association and 9th International Conference on Soft Computing and Intelligent Systems (IFSA-SCIS)

Image inpainting using colour and gradient features

Key Takeaway: Our new inpainting algorithm for color images, using color and gradient features, produces better textures and results compared to those only considering color features.