Text Line Detection and Segmentation: Uneven Skew Angles and Hill-and-Dale Writing

Ergina Kavallieratou (University of the Aegean, Greece)

Fotis Daskas (Greek Open University, Greece)

Abstract: In this paper a line detection and segmentation technique is presented. The proposed technique is an improved version of an older one. The experiments have been performed on the training dataset of the ICDAR 2009 handwriting segmentation contest in order to be able to compare, objectively, the performance of the two techniques. The improvement between the older and newer version is more than 24% while the average extra CPU time cost is less than 200 ms per page.

Keywords: Line detection, OCR, line segmentation

Categories: I.7.5