FALL 1999
This information effective for Fall 1999.
Check with instructor the first day of class for any changes.
The course meets on Tuesdays and Thursdays, from 6 PM to 7:45 PM; or from 6 PM to 7:50 PM if students opt for a 5 minute stretch from 7PM to 7:05PM. Room AS 156. Students *must* attend the first lecture to maintain preregistered place unless prior arrangements made with instructor. Course conduct governed by SOE code of ethics, and attachement to this sylabbus on first day of class. The course uses a textbook: W. Pennebaker and J. Mitchell, JPEG Image Compression, Van Nostrand Reinhold, New York, 1993. It will be supplemented by handouts. Week 1. Sept 23, 1999, Lecture 1. HW 1 distributed. Homework 1 distributed, due April 9. Review of prerequisites as needed for the course. Introduction to image compression fundamentals and international standards. Review of matrix algebra, matrix inverse. Walsh-Hadamard matrices and the Walsh-Hadamard Transform (WHT). A simple 2-point WHT in one-dimension (1D), and its 2D version. Sequency of basis vector defined. If time, video from SPIE tutorial. Baseline JPEG, 8-bit/pixel. Example. Week 2. Sept 28, 30, 1999. HW 2 distributed, due Oct 7. Lectures 2, 3. Two-Dimensional basis matrices. Popularity of even symmetric DCT. Basics of the discrete cosine transform (DCT). Separability: how 1-D transforms on vectors can apply to 2D matrices. Transform coefficients. Orthonormal transform as energy-preserving coordinate rotation. N-point DCT and 4-point example. Fourier series of periodic waveforms in time domain with combinations of sinusoids. Magnitude, phase, frequency domain and spectral frequency. Transforms as "energy-packing" systems; method of principal components and eigensystems. Karhunen-Loeve-Hotelling transform, and relation to the DCT for JPEG. Introduction to Baseline JPEG. Week 3. Oct 5, 7. Oct 7: HW 3 distributed for Oct 14, and HW 2 collected from Oct 20. Lectures 4 and 5. Lecture 4. Baseline JPEG, 8-bit/pixel. Example, possible video. Lecture 5. Baseline JPEG continued. Color systems, Human Vision System. Individual contributions to JPEG compression from quantization, runlength coding, and entropy coding. Color images, color systems, RGB and YUV. Properties of Human Visual System (HVS), Video on HVS. Week 4. Oct 12, 14. Oct 14: collect HW 3, Distribute Hw 4. Lecture 6. Quantization matrices, artifacts, image quality measures. Review of HVS; contrast sensitivity, Modulation Transfer Function, and Spatial Masking. JPEG source of compression. Lecture 7. Study of the markers and technical requirements of the JPEG bitstream. Options beyond baseline JPEG algorithm. Progressive and hierarchical transmission of the coefficients. Week 5. Oct 19, 21. Lecture 8. Additional topics Lecture 9, possible mini-project presentation. Week 6. October 26. Lecture 10. Mini-project presentations.