Handwriting recognition for a digital whiteboard collaboration platform

The research presented in this paper addresses challenges at the intersection of two disciplines: web based collaboration using digital whiteboards and handwriting recognition. The main focus is on the handwriting recognition in order to enable asynchronous usage of the whiteboard content beyond the existing web portal. We present a way to analyze unstructured whiteboard content including drawings, sketches and handwritten text. Our approach uses a recursive extension of the DBSCAN algorithm in order to transfer smaller portions of content to the recognition engine and achieve an appropriate spatial clustering of the content. The adjustment of the configuration parameters, as well as the development of a break condition for the recursion, are shown in detail. We show that it is possible to use an online handwriting recognition engine with offline data and still achieve meaningful results. The presented architecture on the one hand, and the combination of online and offline recognition on the other, ease asynchronous modes of interaction using digital whiteboards.