Information Theory and Coding –
Image, Video and Audio Compression
2002-03
Lecturer: Dr Markus Kuhn
Lectures 2-4 of the Information Theory and Coding
course provide a very brief introduction into practical lossy and
lossless audiovisual compression techniques (2003 January 20, 22 and
24).
Syllabus:
- Perceptual coding versus entropy and channel coding
- Sampling and aliasing
- Quantisation, μ-law coding
- Psychophysics: Weber's law, Fechner scale, Stevens' law
- Decibel
- YCrCb colour representation and chroma subsampling
- Decorrelation of neighbour pixels via Karhunen-Loève and
Discrete Cosine Transform
- JPEG (DCT baseline), G3-fax and JBIG1 image compression standards
- MPEG video compression
- Psychoacoustics: cochlea physiology, sensation thresholds,
spectral and temporal masking (including demonstration)
- MPEG-1 audio compression overview
- voice signals
Study Materials:
- Slides (2up,
4up)
- Additional slides (2up, 4up) with scanned illustrations
from some of the quoted literature [a paper copy of somewhat better
quality than these scanned transparencies is available in the Computer
Lab library on the lecture-notes table near the photocopier]
- Audio files loudness.wav and masking.wav used in the audio demonstration on
January 24 (see slide 27)
Information for supervisors (ask
lecturer for password)
|
