Research On Scanned Documents Direction Correcting And Document Fragments Splicing

Splice the document fragments using computer technology has important application prospect in heritage restoration,criminal forensics and other fields.At present, most of the document splice algorithm treat the document as image, and then use image processing technique to splice them.The author find that the document fragments have notable feature which the text makes trip, whether can we use this feature to look a new algorithm which is different from the mosaic image fragments and play the making trip feature of document to splice them is the focus of this paper.Through screening of large number of image processing algorithms,we find the Mojette transform algorithm which has been applied to image texture recognition is suitable for the feature of the document.Therefore, this paper base on Mojette transform,use line of text in the document as the tilt angle,propose an correction algorithm of the document fragments,which has small calculating amount and good direction selecting features.Base on the above algorithm,use the traditional image edge detection algorithm,execute boundary detection on the document fragments which have been corrected direction.And then use the features which the text make trip, to execute text region extraction, text feature extraction,and ultimately proposee a special stitching algorithm for document fragments.In order to verify the validity of the algorithm,we selecte the most commonly used feature points splicing method which based on SUSAN to carry on comparative experiment.The experimental result validates that the algorithm which proposed in this paper splices much better.