EE381K-14 Multidimensional Digital Signal Processing - General M-D Sampling

Lecture by Prof. Brian L. Evans

Announcements

Image compression standards use rectangular sampling and apply separable systems. The increased complexity for non-rectangular sampling and non-separable systems, even when it is a model of the human visual system, has not provided enough coding gain for increased complexity incurred. This statement gave rise to the following questions in Spring 2003, to which Dr. Faouzi Kossentini (UB Video) answered:
- What happens when a JPEG bitstream is fed into a JPEG 2000 decoder?
  FK: It hangs or crashes.
- How does JPEG 2000 handle floating-point images? (These are common in radar imaging.)
  FK: J2K only handles integers (although the precision can be as large as 128 bytes!!!), to the best of my understanding.
- What image formats are supported by JPEG and JPEG 2000?
  FK: For both JPEG and J2K, only a few formats are supported, but converters are available everywhere.

Lecture

Lecture on General Multidimensional Sampling
- Part I (Theory) is available in Word and PDF formats
- Part II (Special Cases) is available in PDF, Postscript, and HTML formats

Last updated 01/29/08. Send comments to (Mailbox) bevans@ece.utexas.edu