Performance Evaluation of Curled Textlines Segmentation Algorithms

Syed Saqib Bukhari, Faisal Shafait, Thomas Breuel

In: 9th IAPR Workshop on Document Analysis Systems. IAPR International Workshop on Document Analysis Systems (DAS-2010) June 9-11 Boston MA United States online only (DAS Workshop Web page) 6/2010.


Curled textlines segmentation is a necessary initial step for the hand-held camera-captured document image processing. Curled textlines information is often used as an intermediate step for camera-captured document image dewarping. Curled textlines information can also be used for other camera-based document image processing tasks, like layout analysis etc. So far no work has been done for the performance evaluation and the comparison of already reported curled textlines segmentation algorithms on the standard datasets and performance evaluation metrics. In this paper, we use publicly available camera-captured document image dataset and vectorial performance evaluation metrics to compare different algorithms. We present performance evaluation results on previously reported curled textlines segmentation algorithms.


Bukhari-Curled-Textline-Segmentation-DAS10.pdf (pdf, 2 MB )

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence