Handwritten Text Line Structure Detection via Vector-Field Modeling.

Handwritten Text Line Structure Detection via Vector-Field Modeling.

Mestetskiy L.M. (MSU, Moscow, Russia)
Smirnova V.S. (MSU, Moscow, Russia)

Abstract

We propose an original method for line-by-line segmentation of handwritten documents consisting in constructing a vector field of local text orientations and forming centerlines of lines based on integral curves of the constructed field. We develop and implement an algorithm that robustly handles real images of archival documents, including challenging cases with curved lines, non-uniform interline spacing, and handwriting variability. Experimental results demonstrate high effectiveness (F_1 = 0.92).

Keywords

ext line segmentation; handwritten documents; vector field; vector lines; Delaunay triangulation.

Edition

Proceedings of the Institute for System Programming, vol. 38, issue 3, part 3, 2026, pp. 135-148

ISSN 2220-6426 (Online), ISSN 2079-8156 (Print).

DOI: 10.15514/ISPRAS-2026-38(3)-41

For citation

Mestetskiy L.M., Smirnova V.S. Handwritten Text Line Structure Detection via Vector-Field Modeling.. Proceedings of the Institute for System Programming, vol. 38, issue 3, part 3, 2026, pp. 135-148 DOI: 10.15514/ISPRAS-2026-38(3)-41.

Full text of the paper in pdf (in Russian) Back to the contents of the volume