Image Repair and Restoration Using Deep Learning

Abstract

The art of image restoration lies in finding missing information inside the image and returning it back to its original state. While the field of fine arts and photographic restoration has been known and studied for years, recent technological advances have eliminated time-consuming processes and made image restoration far more feasible and easier to do. A novel method known as image inpainting has become a visionary tool for researchers in the field of image processing as digital picture editing has become more adaptable and image processing tools have improved. To understand how this method has become invaluable especially in the field of computer vision, we choose a deep learning technique that accepts multiple damaged photos as input undergoes training to construct a model to recognize missing information. We then validate the performance of the model to identify how well it can restore an image using multiple performance metrics. The process generates a series of successfully remastered images.