Paper Title
Overview of Document Image Denoising and Restoration Techniques

Abstract
Quality enhancement and restoration of degraded document images is a classical image processing problem that aims to reconstruct visually better document images from degraded ones by removing the artifacts. It is the preprocessing stage of the processing pipeline in Optical Character Recognition (OCR) and Text Detection. Research papers dealing with document image denoising vary substantially. This is mainly due to variation in types of noises and dissimilarity in problem-solving approaches adopted to eliminate them. As a result, multiple image denoising algorithms have evolved over time and have been applied to this task of restoration in image processing. This paper presents an overview of such techniques based on motivation and principle behind the approach, with emphasis on deep learning methods. We compare the methods in terms of qualitative analysis and point out strengths and weaknesses of every method. Keywords - Document Image Restoration, Denoising, Deblurring, Binarization, Deep learning, Image Processing.