Determination and restoration of orientation in scanned image of Multicolumn document (PDF)

Research Paper (undergraduate) from the year 2015 in the subject Computer Science - Programming, grade: Post Graduate, Amity University (Amity University), course: B.Tech & M.Tech, language: English, abstract: The first step of OCR to scan a document. In the process of scanning by OCR system the document is taken as row wise to find the lines of test under the given document. After the determination of row wise lines of document the determination of words and characters by scanning column wise for each pixel. Now each obtained graphical symbol of scanned document is matched with the dictionary of symbol. If at the time of scanning the document is not properly align into OCR then symbol which is being obtained cannot be determined by the dictionary. If the document is unaligned into OCR then the image which is being scanned is rotated accordingly as required to assure that the symbol should be determined properly by the dictionary. Here, the skew detection and correction can correct the angle of 60 degree or less then it. Hence, a system of skew detection and correction of more than 60 degree of skew is required. The orientation detection and restoration is very important step for character recognition. There is need for OCR system to determine the orientation of document having 90 degree or above skew. Currently, there is not any system which can detect the orientation of document containing Indian scripts. In this paper I present the ideal method for detecting the orientation of document having more than 90 degree of skew containing Indian script.