Document cleanup using page frame detection

IJDAR(2008) 11:81–96DOI10.1007/s10032-008-0071-7ORIGINAL PAPERDocument cleanup using page frame detectionFaisal Shafait · Joost van Beusekom · Daniel Keysers ·Thomas M. BreuelReceived:12March2008/Revised:1August2008/Accepted:25August2008/Publishedonline:30September2008©Springer-Verlag2008Abstract When a page of a book is scanned or photo- page frame detection in practical applications, we choosecopied,textualnoise(extraneoussymbolsfromtheneighbor- OCR and layout-based document image retrieval as sampleing page) and/or non-textual noise (black borders, speckles, applications. Experiments using a commercial OCR system…)appearalongtheborderofthedocument.Existingdocu- showthatbyremovingcharactersoutsidethecomputedpagementanalysismethodscanhandlenon-textualnoisereason- frame,theOCRerrorrateisreducedfrom4.3to1.7%ontheably well, whereas textual noise still presents a major issue UW-III dataset. The use of page frame detection in layout-for document analysis systems. Textual noise may result in based document image retrieval application decreases theundesired text in optical character recognition (OCR) out- retrievalerrorratesby30%.putthatneedstoberemovedafterwards.Existingdocumentcleanupmethodstrytoexplicitlydetectandremovemarginal Keywords Document analysis · Marginal noise removal ·noise. This paper presents a new perspective for document Documentpre-processingimagecleanupbydetectingthepageframeofthedocument.The goal of page frame detection is to find the ...