Notes

1. INTRODUCTION

Document images usually suffer from several defects mainly due to natural ageing, environmental conditions, usage, poor storage conditions of the original, as well as human manipulations during the scanning process. Since document image analysis and recognition methods usually assume a smooth background and a good quality of printing or writing, it is imperative to have an efficient image enhancement process in order to restore the good quality of document images. Moreover, document image enhancement process enhances the readability of text areas and permits less image storage space.
Google Drive’s Optical Character Recognition (OCR) lets you convert PDF files with images with text into text documents with support for over 200 different languages. The major disadvantage is that it’s limited when it comes to resolving problems mentioned previously, that are obtained from low contrast and uneven background illumination, bleed-through, shining, or shadow-through effects, damaged characters or noisy background, borders or parts of neighboring page, etc.
This paper is organized as follows: section 2 provides the general techniques for document image enhancement and normalization, section 3 will review the current state of art on document image enhancement and normalization applications, section 4 provides an insight to OpenCV functions that can be used before Google’s OCR, and section 5 concludes the paper.