尤加利精油的功效Turbo R ecognition:Decoding P age L ayout
Taku A.Tokuyasu
Computer Science Division,EECS
U.C.
Berkeley
手英语怎么说
(a)(b)
Figure 1:(a)Article title page;(b)TR decoding into title,abstract,and two-column text zones.
婴儿益生菌的作用与功效
之问1.INTRODUCTION
Scanned document images play an important role in dig-ital libraries [1],both for archival purpos and the dis-mination of content online [2].Document image analysis (DIA)eks to unlock the potential of such documents for electronic reu by analyzing physical,symbolic,and logi-cal image structure.Page layout analysis is concerned with physical and logical gmentation of a page ,to extract the header of an article or a table cell containing a stock price.Efficient optimal interpretation of grammar-bad page layout has been an elusive goal [3,4,5].Turbo recognition (TR)[6]us stochastic finite state transducers to efficiently par the structure of an image.It produces results that are nearly statistically optimal [7].
Figure 1illustrates zoning for metadata extraction from a scanned title page using TR.Figure 2show a partial TR result which is nonetheless uful for gmentation of a new-paper clipping.
惠州学院专业
Many extensions of the prent work are envisioned,in-cluding the integration of layout analysis and OCR,and the application to grayscale images.
2.ACKNOWLEDGMENTS
洗马池>李鸿章
I thank Phil Chou and Richard Fateman for valuable dis-cussions.This work was supported by the Berkeley Digital Library Project under NSF grant number CA98-17353.
Copyright is held by the author/owner.
JCDL’01,June 24-28,2001,Roanoke,Virginia,USA.ACM
1-58113-345-6/01/0006.
(a)(b)
Figure 2:(a)Newspaper clipping;(b)TR decoding (not fully converged).
3.REFERENCES
[1]UC Berkeley Digital Library Project.Part of the NSF
Digital Libraries Initiative.elib.cs.berkeley.edu.[2]R.Phelps,T.A.and Wilensky.Multivalent documents.
Communications of the ACM ,43(6):82–90,June 2000.[3]Alan Conway.Page grammars and page par
sing.a
狐狸狐狸syntactic approach to document layout recognition.In Proc.of 2nd International Conference on Document Analysis and Recognition (ICDAR ’93),pages 761–4,Los Alamitos,CA,1993.IEEE Comput.Soc.Press.[4]P.A.Chou.Recognition of equations using a
two-dimensional stochastic context-free grammar.In Visual Communications and Image Processing IV,Proc.of the SPIE ,volume 1199,pages pt.2:852–63,1989.
[5]P.A.Chou and G.E.Kopec.A stochastic attribute
grammar model of document production and its u in document image decoding.In Document Recognition II,Proc.of the SPIE ,volume 2422,pages 66–73,1995.[6]Taku A.Tokuyasu and Philip A.Chou.An iterative
approach to document image analysis.In DLIA99workshop,Bangalore,India ,1999.
www.science.uva.nl/events/dlia99/.[7]Taku A.Tokuyasu and Philip A.Chou.Turbo
recognition:a statistical approach to layout analysis.In Document Recognition VIII,Proceedings of the SPIE ,volume 4307,pages 123–129,2001.
