Abstract | ||
---|---|---|
In this paper we present a novel model based approach todetect severely broken parallel lines in noisy textual documents.It is important to detect and remove these lines so thetext can be segmented and recognized. We use DirectionalSingle-Connected Chain, a vectorization based algorithm,to extract the line segments. We then instantiate a parallelline model with three parameters: the skew angle, the verticalline gap, and the vertical translation. A coarse-to-fineapproach is used to improve the estimation accuracy. Fromthe model we can incorporate the high level contextual informationto enhance detection results even when lines areseverely broken. Our experimental results show our methodcan detect 94% of the lines in our database with 168 noisyArabic document images. |
Year | DOI | Venue |
---|---|---|
2003 | 10.1109/ICDAR.2003.1227625 | ICDAR-1 |
Keywords | Field | DocType |
estimation accuracy,noisy textual document,detection result,high level contextual,novel model,line segment,model-based line detection algorithm,parallelline model,fromthe model,approach todetect,image segmentation,image recognition,image processing,optical character recognition,text analysis | Line segment,Computer science,Image processing,Vectorization (mathematics),Image segmentation,Artificial intelligence,Computer vision,Vertical translation,Contextual information,Pattern recognition,Optical character recognition,Algorithm,Parallel | Conference |
ISSN | ISBN | Citations |
1520-5363 | 0-7695-1960-1 | 7 |
PageRank | References | Authors |
0.49 | 10 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Yefeng Zheng | 1 | 1391 | 114.67 |
Huiping Li | 2 | 574 | 38.28 |
David Doermann | 3 | 4313 | 312.70 |