Multi-Oriented Text lines Detection and Their Skew Estimation

There are some documents where text lines are not parallel to each other i.e. different text lines of a single page may have different inclinations (orientations) with the horizontal lines. To enhance the ability of document analysis system, we need text line extraction in multiple orientations. In this papers we propose a robust technique (a) to detect text lines of arbitrary orientation in a single document page, and (b) to detect skew angle of individual text line. We use here a bottom-up approach where the connected components are at first labeled. They are then clustered into word groups. Text lines of arbitrary orientation are segmented from the estimation of these word groups. From an experiment of 3045 text lines we obtained an accuracy of 97.7% by the proposed method.