|
Term 1
Term 2
Term 1
Week 1
Summary of Work Done
- Created web-page for project.
- Searched for material on WWW. Minutes for
Meeting on Friday 12th October 2001
- Defined project - decided to focus on lip-synch problem of facial animation.
- Defined 3 main sections of project - mouth movement (and speech) data capture
analysis of mouth movement and phonemes
application of results to facial animation.
- Decided on mouth movement data capture through video capture card and camera.
- Arranged to install hardware and software on Monday 15th October. Week
2 Summary of Work Done
- Installed video capture card and camera, along with software needed to
use it.
- Studied existing video camera code.
- Acquired reading material on computer vision.
- Worked on C programming. Minutes for Meeting on Monday 15th October 2001
- Installed hardware and software
- Overview of usage of card and possible application methods for project.
- Looked at speech analysis software. Minutes for Meeting on
Friday 19th October 2001
- Discussed progress so far
- Overview of entire project, and how everything fits together.
- Considered possible methods of capture - computer vision
- Considered possible formats of data capture - e.g. y-axis displacements of
points on mouth.
- Discussed how data captured would be applied to facial model -through
generation of functions from analysis of data. Week
3 Summary of Work Done
- A bit
more work on C programming.
- Began
learning to do Windows C Programming.
- Experimented with
camera to save clips as AVI files.
- Read a bit on vision from textbooks. Minutes for Meeting on
Friday 26th October 2001
- Decided to abandon working with Vision SDK – it doesn’t have the required
functionality.
- Decided to use Direct Show to manipulate video clips.
- Discussed use of a non-reflective marker to be stuck on mouth for tracking.
- Discussed problems with saving video clips as AVI files
– compression time is quite long. Week
4 Summary of Work Done
- Did some more work on Windows C Programming.
- Began learning how to use DirectX and DirectShow.
- Did some preliminary DirectX and DirectShow programming - managed to play back
a file.
- Found a possible method of grabbing a frame using DirectShow. Minutes for Meeting on
Friday 2nd November 2001
- Went over what I had done during the week - using DirectShow.
- Discussed use of the 'grab a frame' method.
- Decided on being able to visually see the modifications made to an AVI file by
playing it back with the modifications perhaps overlaid on the AVI being played
back, or by modifying the extracted frame and displaying it.
- Found a solution to the sound extraction from the AVI file.
- Plan for next week - read in a file and play it back with modifications on the
screen. Week 5 Summary of Work Done
- Managed to use DirectShow to grab a frame.
- Using a loop, was able to play back every single grabbed frame by displaying
each frame as a bitmap.
- Managed to modify the bitmap and change the 'rgb' values of pixels - modified
image was displayed on the screen.
- Problems - the DirectShow graph plays back the clip in a separate window in
addition to the bitmap display. Minutes for Meeting on
Monday 12th November 2001
- Showed progress so far with DirectShow.
- Discussed getting started with vision algorithms - create filters for mouth
tracking.
- Discussed DirectShow 'window' problem (clip plays back in another window). Week
6 Summary of Work Done
- Acquired reading material on vision algorithms
- Did some preliminary reading.
- Continued to search for DirectShow 'window' problem. Minutes for Meeting on
Friday 16th November 2001
- Discussed possible vision algorithms to use as filters. Week
7 Summary of Work Done
- Applied a simple vision filter that uses normalized RGB values, decides if
they fit within a (controllable) range, and modifies the output display if it
does.
- Fixed the DirectShow 'window' problem.
- Determined how to stop the processing once the end of the clip has been
reached. Minutes for Meeting on Monday 26th November 2001
- Demonstrated the simple filter and progress so far.
- Discussed using a histogram based filter. Week
8 Summary of Work Done
- Attempted to implement a histogram based filter.
- Problems with implementing it - problem with scaling down normalized values to
a range from 0 - 16. Week 9 Summary of Work Done
- No work done - EPM Assessment. Week
10 Summary of Work Done
- No work done - EPM Assessment / End of term. Term
2 Week 1 Summary of Work Done
- No work done - Exams. Week
2 Summary of Work Done
- Cleaned up and organized code.
- Optimized code.
- Fixed DShow bugs. Minutes for Meeting on
Friday 18th January 2002
- Discussed progress so far.
- Plan for rest of term - cut back scope of project - no implementation of
results.
- Aim to achieve comparison between mouth positions and phoneme occurrences.
- Plan report layout. Week
3 Summary of Work Done
- Improved histogram algorithm by adding a 'don't detect' zone.
- Experimented with different clips using different lighting conditions to find
good recording conditions.
- Made a rough report layout plan.
- Ordered coloured lipstick to improve tracking. Minutes for
Meeting on Friday 25th January 2002
- Discussed work done over the week.
- Need to improve the report layout plan - more detail.
- Discussed methods of storing/generating output from filter - curve fitting
representation of lip.
- Discussed contour tracking.
- Application of results to 'Linguistic Sound Editor' output to perform some
sort of comparison. Week
4 Summary of Work Done
- Downloaded and used Microsoft Linguistic Sound Editor.
- Tried different camera and lighting conditions - little improvement.
- Applied erosion and dilation algorithms - there was some improvement.
- Received new lipstick. |