Identification of the Global Text Skew Based on the Convex Hulls

Old printed documents represent the important part of our heritage. For theirs preservation, the digitalization is indispensable. The paper proposes the algorithm for the identification of the global text skew. It is based on the connected components made by the filled convex hulls around each text element. Furthermore, connected components are enlarged by morphological dilatation. Currently, the biggest connected component is extracted. The global orientation of the document is identified by its orientation. Efficiency and correctness of the algorithm is examined by testing on datasets. Obtained results proved the algorithms correctness.