Seite - 18 - in Document Image Processing
Bild der Seite - 18 -
Text der Seite - 18 -
J. Imaging 2018,4, 27
capableoffiltering-outsuchanoise increasesenormously,asanewsetofhuesofpaperandprinting
colorsappear. Thedirectapplicationofbinarizationalgorithmsmayyieldacompletelyunreadable
document, as the interfering inkof the backside of thepaper overlapswith the binary one in the
foreground. Severaldocument imagecompressionschemes forcolor imagesarebasedon“adding
color” toabinary image. Suchcompressionstrategy isunable tohandledocumentswithback-to-front
interference [5]. OpticalCharacterRecognizers (OCRs) are alsounable toworkproperly for such
documents. Severalalgorithmsweredevelopedspecifically tobinarizedocumentswithback-to-front
interference [3,4,6–9]. There isnobinarizationtechniquetobeanall casewinnerasmanyparameters
mayinterfere inthequalityof theresultingimage[9]. Thedevelopmentofnewbinarizationalgorithms
is still an important research topic. International competitionsonbinarizationalgorithms, suchas
DIBCO-Document ImageBinarizationCompetition [10], areanevidenceof therelevanceof thisarea.
Figure1. Imageswithback-to-front interference fromthe three test setsused in thispaper:Nabuco
bequest (left),LiveMemory(center) andDIBCO(right).
This paper presents a new global filter [1] to binarize documents, which is able to remove
the back-to-front noise in awide range of documents. Quantitative and qualitative assessments
made inawidevarietyofdocuments fromthreedifferent“real-world”datasets (typed,printedand
handwritten, using different kinds of paper, ink, etc.) allow to witness the efficiency of the
proposedscheme.
2.TheNewAlgorithm
Thealgorithmproposedhere isperformedinfoursteps: 1. decision-makingforfindingthevector
ofparametersof the imagetobefiltered,2. filteringthe imageusingabilateralfilter,3. splittingthe
image into theRGBcomponents, andperformingtheirbinarizationusingamethodinspiredbyOtsu’s
algorithm for eachRGB channel, and 4. choice ofwhich of theRGB components best preserved
thedocument information in the foreground,which isconsideredthefinaloutputof thealgorithm.
Figure 2presents theblockdiagramof theproposedalgorithm. The functionalityof eachblock is
detailedas follows.
18
zurück zum
Buch Document Image Processing"
Document Image Processing
- Titel
- Document Image Processing
- Autoren
- Ergina Kavallieratou
- Laurence Likforman-Sulem
- Herausgeber
- MDPI
- Ort
- Basel
- Datum
- 2018
- Sprache
- deutsch
- Lizenz
- CC BY-NC-ND 4.0
- ISBN
- 978-3-03897-106-1
- Abmessungen
- 17.0 x 24.4 cm
- Seiten
- 216
- Schlagwörter
- document image processing, preprocessing, binarizationl, text-line segmentation, handwriting recognition, indic/arabic/asian script, OCR, Video OCR, word spotting, retrieval, document datasets, performance evaluation, document annotation tools
- Kategorie
- Informatik