Online Handwriting Analysis and Recognition
This chapter presents methods and results towards the analysis and recognition of handwritten notes taken during meetings that are held in smart meeting rooms (see Chapter 1), including two types of data: whiteboard notes and notes taken with a ballpoint pen. We will focus on three topics which are important to transform handwritten notes into a machine readable format. First, the detection of the drawing mode, either handwriting or drawn graphics, is investigated in order to retrieve the handwritten content. Second, we present methods for online handwriting recognition. Third, we discuss the problem of writer identification, i.e., detecting which person wrote a given text. In all three domains, the novel approaches described here outperform previous approaches and in some cases even commercial recognizers.