A Design And Implement On Image Restoration System For Pdf Document

With the rapid development of information technology and Internet, people can easily obtain the required electronic data which are stored in computers as documents, and a more widely used document format is PDF. However, due to the reason of converting form paper-based materials to electronic data, there are often phenomena of writing broken and illegibility in the PDF documents. In this thesis, to solve this problem and to promote the readability, the PDF documents are processed by image restoration and enhancement technology.The study in this thesis includes the following sections:1. The first section focuses on the analysis of the cause of fouling marks in PDF documents, the MADM method for detecting noise was introduced, and the opening and closing filter method for de-noisy was designed.2. Based on imaging principle of scanner, the cause of motion blur was analyzed in the next part. Then based on scanning imaging feature and motion-blur restoration method, a blur-image restoration algorithm is designed to process blur-images obtained from scanning.3. The differences between the words in scanning image and in manually typing were analyzed. The cause of the illegible writing was explored and the slide-window anisotropic interpolation algorithm was designed for this situation.At last, the PDF document restoration system was realized by C++language with Visual Studio.Net integrated development environment and OpenCV library. With this restoration system, five kinds of PDF documents can be restored, including fouling marks in PDF document, scanning tilt, non-sharp image, blur image and illegible writing. The results show that the restored images were satisfactory and more readable than fouling images.