My current main interest is in software for Document Image Analysis (DIA), and/or support for Rich GUI applications in the cloud.
DIA is about analysing images of documents of various types and turning them into semantically meaningful knowledge. There are many potential projects in this area, from the deeply image analysis (e.g. image cleaning, deskewing etc.) to optical character recognition (glyph identification, glyph segmentation, pattern classification, shape clustering), to domain specific (analysis of music sheets, mathematical formulae, technical diagrams, maps etc.) to sophisticated algorithms (e.g graph grammar parsing) to AI-like knowledge construction (e.g. term and graph rewriting, mathematical knowledge base).
Rich GUI applications in the cloud refers to building web applications with a focus on high performance and sophisticated graphical user interfaces. The arrival of HTML5 with its new Canvas element, and especially when combined with Javascript and CSS, gives web applications the ability to build interfaces nearly as powerful and flexible as available from Java Swing. This opens the possibility of avoiding the problems of distribution, porting, and version management of traditional GUI applications and instead simply providing the application on the web. I am very interested in exploring these technologies to see how rich an experience we can give users in such web applications. Most of my projects that are GUI front end oriented can be taken as either a desktop application or as a rich GUI web application.
Overall I tend to like the more technically demanding types of project where there are interesting implementation issues or serious problems of design, analysis or mathematics to deal with. Some of my projects have a strong research element and a good result from the project may lead to publication in international conferences or journals. There are also opportunities to pursue some of these projects further into PhD research projects following graduation.
While these projects can be developed in Java, I have switched to using OCaml in my own work for developing code in this area because of the improved speed of development, speed of execution and because I find it easier to write complex algorithms in this language. If you would like to use OCaml for the project (and I heartily recommend doing so) you must either be an MSc or undergraduate student with some previous familiarity with OCaml, or an undergraduate student who is willing to learn OCaml over the Summer with support from me.
The general subject of Document Image Analysis covers everything from low level image analysis up to high level sophisticated algorithms for turning components identified in an image into usable information. I am particularly interested in developing a centre of expertise in Document Image Analysis here at the University of Birmingham. There are many related projects available here and I will generally give preference to DIA projects over other projects. There is plenty of scope for interested students to pursue this topic further at the MSc level (via our MSc in Advanced Computer Science) and at PhD level. A number of previous projects in this area led to papers being published. In one, in 1998/1999. the results were subsequently commercialised and it is now a live font search engine run by a major font foundry in the USA. See Identifont on http://www.myfonts.com. The student involved has been, and remains, employed by that company.
OCaml is a strongly typed, high performance programming language with functional imperative and object oriented features as well as a type inference system (so although strongly typed, you often need not explicitly declare the types you are using). Its features (such as pattern matching, and type inferencing) makes much simpler the process of constructing and manipulating complicated structures of data than is the case in languages such as Java, C, C++ or many other languages. I am interested in supervising some projects to explore the use of this language on some complex problems.
The following is a list of projects intended to explore a number of aspects of Web Application Technologies.
I am interested in the possible connections between music, sound and computers. From analysing music from sound files to extract the more abstract music information (e.g. producing a music sheet from the sound file - an open research problem) to doing the same from MIDI data (much easier but still not simple), to optical recognition of sheet music to produce MIDI data (i.e., teaching your computer to read music) etc.
The following projects are to do with tools that assist developers to design and build software systems. I am also interested in other projects of this type if you have any ideas for something related.
Maintained by A.P.Sexton@cs.bham.ac.uk
Home Page: http//www.cs.bham.ac.uk/~aps
School of Computer Science Home Page