1. Introduction
Optical character recognition (OCR) is an important step in the process of digitizing important documents such as old historic books. OCR research over the last few decades has led to highly accurate digitization of documents. However there is a severe drop in the performance of OCR systems in the presence of distortion (warping) in the scanned/photographed document image as shown in Fig. 1. These systems rely on the document image being planar and having straight horizontal text lines. Therefore, it is critical to remove any distortion that might exist in the document image.
Examples of distortions: (a) Distortion at book bindings, (b) Perspective projection in camera captured image. (c) One of the applications of proposed method: Mobile document scanner (image from google).