Recognising drawn symbols

Software	Java or C++ or Smalltalk or another object oriented language with good GUI facilities
Covers	Algorithm design, graphical user interfaces
Skills Required	Programming, some interest in graphics and/or artificial intelligence
Challenge	Conceptual ??? Technical ?? Programming ????

Brief Description

A major part of increasing the power of graphical user interfaces is enabling users to provide input in ways that are more efficient and powerful than selecting icons or memory items with a mouse. One vital form of input is handwriting or drawing. Although understanding handwriting is a very complex problem, recognising isolated symbols from a sequence of screen coordinates is relatively tractable. Handwriting software is becoming increasingly effective and widespread in handheld computers. However what applications developers would need is a freeware class library to provide drawn symbol recognition capabilities within their own applications.

The objective of this project is to develop a set of techniques for recognising symbols or glyphs drawn on a screen, and implement them as a well written and well documented library of methods in the chosen programming language, that can be used in real life applications (for example by another student doing a BSc project, who wants to use drawn symbol input).

Recognising drawn symbols is an open-ended problem that can be tackled in a number of ways. A major part of the project is thinking up and testing ways of identifying features of drawn symbols.

Extensions

A system that allows the user to define a set of symbols to be recognised in some reasonably user-friendly way.

A system that learns different patterns from training input provided by the user.

Variants

A system that enables users to draw data flow diagrams (or some other type of graphical notation), just using the keyboard to type label names.

Back to