Extraction of Original Text Document from a Set of Degraded Text Documents from the Same Source

Information extraction is the task of extracting structured data from a degraded document. It includes data extraction such as text, image or graphics from the sources such as an image, video or documents. Text detection and extraction from the degraded document finds application in wide range of study. In this paper, an Optical Character Recognition less (OCR-less) method of obtaining an original text document from a set of degraded text documents is proposed. It involves detection and extraction of text from a set of degraded text documents which belongs to the same source. The degraded text documents are treated as images and are input to our proposed algorithm. It involves Image Processing techniques such as image subtraction and image fusing. Deskewing of input degraded text document images are also proposed by detecting Speeded Up Robust Features (SURF) from the images. A methodology to convert Grayscale image format to RGB image format is also developed using YCbCr image format to obtain an original document in the RGB format same as the input degraded RGB text document images.