A Text Detection and Orientation Restoration System for Mobile Devices

This paper proposes an automated system to detect text regions and correct the orientation of textlines in natural images on mobile devices. Text regions are first detected using an image operator called the Stroke Width Transform. Textlines (text blocks) are then extracted from text regions based on a modified minimum spanning tree algorithm. Finally, the orientations of textlines are corrected according to the way users hold the device. The system provides the users with an interactive interface to select the text blocks which catch their attention. Moreover, the interactive interface allows the users to correct the orientation of the selected text blocks if the results of the automatic correction are wrong. With such a design, wrongly detected text regions or an incorrect orientation of textlines would not confuse the users. The system is implemented on the Android platform and the experimental results demonstrate the feasibility and validity of the proposed system.