Free-Hand Sketch Recognizer for Graphs

My B.Sc. Engineering Project was on Intelligent Human-Computer Interfaces. I have developed a Free-Hand Sketch Recognizer for Graphs in cooperation with Dr. T. Metin Sezgin from Design Rationale Group in CSAIL @ Massachusetts Institute of Technology. Project was supervised by Assist. Prof. Ender Ozcan. The details of the developed system can be found below:

The current trend on the use of touch-screens has introduced the need for keyboard-less solutions to graphical user interfaces. Gesture-based interfaces as a piece of pie, receive the scientific interest because of its theoretical diversity. Even näive approaches require a common knowledge about linear algebra, artificial intelligence, sketch and handwriting recognition within the context. The main emphasis of this project is to develop an intelligent pen-based human-computer interface and propose a new online hand-drawn digit recognition algorithm which runs with linear time complexity, where one of the most recognized algorithms, ICP (Iterative Closest Point) halts with quadratic time complexity. In return, we show how the proposed algorithm, together with ICP, can be applied to increase the success ratio of digit recognition significantly without the inclusion of a neural network. Moreover, the derived information is represented by an adjacency matrix to supply an abstract view of any kind of input graph. As sample applications, Kruskal and Dijkstra algorithms are fed using the output generated by the implemented sketch recognizer system. In conclusion, this work outlines a new solution to the graph creation problem with hand-drawn sketch and digit recognition.

Dibeklioglu, H., "A Sketch Recognizer for Graphs," B.Sc. Engineering Project Report, Yeditepe University, 2006.
[ Download PDF ]

Dibeklioglu, H., T.M. Sezgin, and E. Ozcan, "A Recognizer for Free-Hand Graph Drawings," International Workshop on Pen-Based Learning Technologies (PLT), Catania, Italy, 2007.
[ Download PDF ] [ Download Extended PDF ]

Demo Videos:

Manual Mode Dijkstra Demo [ Download AVI ]
Recognition Mode Dijkstra Demo [ Download AVI ]
Manual Mode Kruskal Demo [ Download AVI ]
Recognition Mode Kruskal Demo [ Download AVI ]