Doctor of Philosophy
Dr. Stephen M. Watt
Pen input for computing devices is now widespread, providing a promising interaction mechanism for many purposes. Nevertheless, the diverse nature of digital ink and varied application domains still present many challenges. First, the sampling rate and resolution of pen-based devices keep improving, making input data more costly to process and store. At the same time, existing applications typically record digital ink either in proprietary formats, which are restricted to single platforms and consequently lack portability, or simply as images, which lose important information. Moreover, in certain domains such as mathematics, current systems are now achieving good recognition rates on individual symbols, in general recognition of complete expressions remains a problem due to the absence of an effective method that can reliably identify the spatial relationships among symbols. Last, but not least, existing digital ink collaboration tools are platform-dependent and typically allow only one input method to be used at a time. Together with the absence of recognition, this has placed significant limitations on what can be done.
In this thesis, we investigate these issues and make contributions to each. We first present an algorithm that can accurately approximate a digital ink curve by selecting a certain subset of points from the original trace. This allows a compact representation of digital ink for efficient processing and storage. We then describe an algorithm that can automatically identify certain important features in handwritten symbols. Identifying the features can help us solve a number of problems such as improving two-dimensional mathematical recognition. Last, we present a framework for multi-user online collaboration in a pen-based and graphical environment. This framework is portable across multiple platforms and allows multimodal interactions in collaborative sessions. To demonstrate our ideas, we present InkChat, a whiteboard application, which can be used to conduct collaborative sessions on a shared canvas. It allows participants to use voice and digital ink independently and simultaneously, which has been found useful in remote collaboration.
Hu, Rui, "Representation, Recognition and Collaboration with Digital Ink" (2013). Electronic Thesis and Dissertation Repository. 1771.