3rd year projects 2004 - 05

  • JPB-UG-1: Snooker Highlights (Matthew Winchombe)
  • JPB-UG-2: Video-Based Snooker Shot Indexing (Guy Philip)
  • JPB-UG-1: Snooker Break Speed Estimation (Robert Walker)
  • JPB-UG-4: Audio-Based Snooker Shot Indexing
  • JPB-UG-5: Gesture Recognition for 'Hands-Free' Human-Computer Interaction (Philip Moore)

Mail all

The project descriptions below are only intended as starting points. If you wish to discuss possibilities in greater detail I encourage you to email me to arrange a meeting.


JPB-UG-1: Snooker Highlights

Description

Snooker is a fairly slow paced game and watching it on TV in ‘real time’ can be rather tedious. Imagine an application that could take a live broadcast of, say, a 30 minute game and reduce it down to 5 minutes of highlights. The shorter version would still show all the shots, but it would omit the boring bits in between that are shown while the players are busy preparing for the next shot.

This project will use video processing techniques to analyse a collection of recorded snooker games that were originally broadcast live. The system will analyse the video to partition it into a sequence of video shots. Then by using computer vision techniques the system will prune out shots that are not wanted in the shortened version (e.g. shots of the audience, close-ups of the opponent waiting in the chair, shots of the player walking around the table etc).

The project will employ recordings of this year’s Embassy World Snooker Championships which will be provided on CD or DVD.

This project will require good programming skills. The project may make use of the <a href=http://www.intel.com/research/mrl/research/opencv/>OpenCV</a> computer vision library. Knowledge of C/C++ (or the willingness to learn) will be an advantage.

Requirements

  • good java or C/C++ programming skills

Reading

[TOP]


JPB-UG-2: Video-Based Snooker Shot Indexing

Description

This project takes a different approach to improving the watchability of snooker. Rather than creating highlights, the application will insert index points to mark the position of each shot in the game. The user will then be able to quickly step through the game by jumping from one shot to the next.

The project will use computer vision techniques to identify the beginning of each shot. Selecting features that can be used to reliably identify these events will be one of the challenges of the project. For example, most shots are preluded by a close-up of the cue ball in which the cue tip is also visible - so looking for frames that contain a large white circle might be a useful starting point.

The project will employ recordings of this year’s Embassy World Snooker Championships which will be provided on CD or DVD.

This project will require good programming skills. The project may make use of the <a href=http://www.intel.com/research/mrl/research/opencv/>OpenCV</a> computer vision library. Knowledge of C/C++ (or the willingness to learn) will be an advantage.

Requirements

  • good java or C/C++ programming skills

Reading

[TOP]


JPB-UG-1: Snooker Break Speed Estimation

Description

This project would use video analysis and tracking techniques to estimate the power of the break (the opening shot) in a game of snooker. Snooker balls are relatively easy to track - they move in roughly straight lines, they have a uniform colour and they move against an approximately uniformly coloured background. By tracking the cue ball during the break it should be possible estimate its actual trajectory on the table. Given its position in each frame it is trivial to calculate the speed of the ball and hence the power of the shot.

The project could be generalised to measure the speed of the cue ball in every shot. One can even imagine building profiles for various players (e.g. It could be shown that Alex ‘Hurricane’ Higgins plays a greater number of power shots than Steve ‘Boring’ Davis) . The techniques developed could also be applied to other ‘slow moving’ ball sports such pool, bowls, curling?? The project would have endless application for sports commentary, e.g. ‘That pot measured 25 kilometres per hour - the fastest shot of the tournament so far!’ ;)

The project will employ recordings of this year’s Embassy World Snooker Championships which will be provided on CD or DVD.

This project will require good programming skills. The project may make use of the <a href=http://www.intel.com/research/mrl/research/opencv/>OpenCV</a> computer vision library. Knowledge of C/C++ (or the willingness to learn) will be an advantage.

Requirements

  • Good java or C/C++ programming skills, good maths

Reading

  • worldsnooker.com - Lots of general information about the game.</li>
  • Gonzales and Woods, Digital Image Processing, Addison-Wesley Pub. Co, Reading, Massachusetts, 1992. (or any other similar textbook)
  • Smith Reviews of Optic Flow, Motion Segmentation, Edge finding and Corner Finding Technical Report TR97SMS1 - available here</li>
  • QuesTec: Sports Technology - real applications.
[TOP]


JPB-UG-4: Audio-Based Snooker Shot Indexing

Description

This project is related to JB-2 but rather than using the video signal it aims to achieve the same goal using the audio track. Again, the aim is to insert an index point into the video corresponding to each shot played in the game. When the shot is played there is a characteristic sound as the cue hits the cue ball. So the project will attempt to use signal detection theory and techniques to locate all occurrences of this sound in the audio track.

The project will employ recordings of this year’s Embassy World Snooker Championships which will be provided on CD or DVD.

Requirements

  • good programming skills

Reading

  • worldsnooker.com - Lots of general information about the game.</li>
  • Dufaux et al. “Automatic sound detection and recognition for noisy environment” Proc EUSIPCO 2002 </li>
  • Whalen, “Detection of signals in noise”, Academic Press, 1971 (St. Georges Library)
  • Poor, “An introduction to signal detection and estimation”, Springer-Verlag, 1988
[TOP]


JPB-UG-5: Gesture Recognition for 'Hands-Free' Human-Computer Interaction

Description

This project would use video analysis and tracking techniques to track a user’s hand across a ‘field of view’ supplied by a digital camera. Combining the tracking with gesture recognition techniques would allow the user to interact with the computer by pointing at objects shown on screen, and by use of predetermined gestures, manipulate these objects, for example, indicating the object be ‘opened’, moved, or to initialise commands upon the object. This will effectively replace input devices such as the mouse.

Such systems already exist with technologies such as the RAVE tool, but the use of a purely optical technique would allow the user to interact with the machine with no actual physical contact. Other than the benefit of no external, moving parts, this may also prove useful for people with disabilities, such as an inability to grip things.

[TOP]