doi:10.1016/S0262-8856(02)00091-4
Copyright © 2002 Published by Elsevier Science B.V. All rights reserved.
Skew angle estimation for printed and handwritten documents using the Wigner–Ville distribution
E. Kavallieratou
,
, N. Fakotakis and G. Kokkinakis
Wire Communications Laboratory, Dept. of Electrical Computer Engineering, University of Patras, 26500, Patras, Greece
Available online 6 September 2002.
References and further reading may be available for this article. To view references and further reading you must
purchase this article.
Abstract
A skew estimation algorithm for printed and handwritten documents, based on the document's horizontal projection profile and its Wigner–Ville distribution, is presented. The proposed algorithm is able to correct skew angles that range between −89 and +89° detecting the right oriented position of the page by the alternations of the horizontal projection profile. It is able of processing successfully handwritten documents, even if they consist of non-parallel text lines. It deals with the presence of graphics, while a few text lines suffice for the application of the algorithm. Furthermore, the latter permits the use of only a part of the page for the skew estimation minimizing the computational complexity. The proposed algorithm was evaluated on a wide variety of pages (i.e. printed, handwritten, multi-column, application forms etc.) achieving a success rate of 100% within a confidence range of ±0.3°.
Author Keywords: Skew angle correction; Skew angle detection; Wigner–Ville distribution
Fig. 1. A handwritten page with non-parallel text lines.
Fig. 2. A skewed application form. Such pages with various fonts, both printed and handwritten, and borders usually complicate the skew angle estimation.
Fig. 4. Horizontal histograms of the document page of Fig. 3 for various skew angles.
Fig. 3. A right-oriented document image.
Fig. 5. Maximum peaks of the histograms referred to the document page of Fig. 3, with respect to the skew angle.
Fig. 6. The WVDs for several histograms of the document page of Fig. 3.
Fig. 8. The proposed algorithm.
Fig. 7. The relation between big step and rotations. The required rotations are minimized for a big step equal to 12°.
Fig. 10. The procedure for the handwritten document of Fig. 9(a).
Fig. 9. (a) A skewed handwritten document, (b) after the first estimation, rotated by 60° and (c) after the final estimation rotated by 62.0°.
Fig. 11. Curves of maximum intensity extracted from Fig. 3.
Fig. 12. Number of times that several curves of maximum intensity for different skew angles present maximum.
Fig. 13. A two-column skewed document.
Fig. 14. The horizontal (a) and vertical (b) histogram of the page of Fig. 13.
Fig. 15. The selected part (a) before and (b) after the correction.
Fig. 16. The average maximum intensity curve, the selected threshold and the extracted areas (A, B, and C) of different skew angle for the page of Fig. 1.
Fig. 17. Several curves of maximum intensity for the page of Fig. 1.
Fig. 18. The corrected page of Fig. 1.
Fig. 19. The confidence range of the estimated skew angle vs. the achieved accuracy.
Fig. 20. A skewed document page with a resolution of 100 dpi (a) before and (b) after the correction.
Fig. 21. Skewed page of application form with a resolution of 100 dpi (a) before and (b) after the correction.
Fig. 22. A skewed page of two-column document with a resolution of 100 dpi (a) before and (b) after the correction.
Fig. 23. A skewed page of handwritten document (a) before and (b) after the correction.
Table 1. Experimental data: document images extracted from PS form

Table 2. Performance example for some of the pages shown in this paper
