A unified framework for document image restoration

Abstract

Document image processing and analysis has been an active research topic in recent years, which includes text detection and extraction, normalization, enhancement, recognition and their related applications. The work described in this thesis focuses on the normalization of various types of document images that display all sorts of distortions including shadings, shadows, background noise, perspective and geometric distortions. In particular, a unified framework is developed which takes in an input image and rectifies all the distortions at one go to produce a final image that facilitates human perception and subsequent document image analysis tasks. The whole framework consists of three main components: photometric correction, surface shape reconstruction, and geometric correction. The first component is designed to address distortions including shadings, shadows and background noise through an inpainting-based procedure. The second component is meant to derive the 3D geometry of the document for the succeeding perspective and geometric correction tasks. It comprises of two Shape-from-Shading methods with different solving schemes for the image irradiance equation formulated under various illumination conditions. Finally, the last component is targeted at perspective and geometric distortions with three proposed methods handling different types of images by utilizing different sets of input information. Results on synthetic and real document images demonstrate that each type of the distortions can be effectively corrected using a full or sub set of the procedures in the whole framework. OCR results on the restored images of those text-dominant documents also show great improvements over the original distorted images.