Abstract: Old printed documents represent the significant part of our heritage. In order to preserve them, the digitalization is indispensable. The paper proposed a robust skew estimation method for old printed document. It is based on the connected components made by filled convex hulls around text element. The connected components are enlarged by oriented morphological operation. Then, the longest connected component is extracted. The global orientation of the document is detected by its orientation. Accordingly, document image was globally de-skewed. The algorithm is tested on synthetic and real datasets. Obtained results proved the algorithmscorrectness.
Loading