Image inpainting with prioritizing of hole's pixels

摘要

Image inpainting is one of the most widely used image processing techniques which removes an object from a picture or reconstructs images with lost parts. A method is presented in this article that takes a picture and the location that should be inpainted as the inputs and repair will be done automatically. To inpaint an image, first we obtain the border of holes, then we calculate the priority of border pixels and according to these priorities, the holes will be filled using the values and gradients of neighboring pixels. The proposed Algorithm runs as fast as image inpainting methods such as FMM and NS which restore images within a few seconds. This algorithm has been evaluated on sample images and satisfactory results have been achieved.