Abstract
Summary form only given, as follows. An extension of Breiman, Friedman, Olshen, and Stone's algorithm for optimal pruning in tree-structured classification and regression is applied to bit allocation. It uses a simple tree model to deallocate bits from classes of a coder, resulting in a set of variable-rate codes of different average rates. When applied to classified vector quantization, it leads to a 3.4-dB gain in the signal-to-noise ratio of a magnetic resonance image coded at 1 b/pixel over fixed-rate classified vector quantization at the same rate.
Original language | English |
---|---|
Pages | 12-13 |
Number of pages | 2 |
State | Published - 1990 |
Event | 1990 IEEE International Symposium on Information Theory - San Diego, CA, USA Duration: 14 Jan 1990 → 19 Jan 1990 |
Conference
Conference | 1990 IEEE International Symposium on Information Theory |
---|---|
City | San Diego, CA, USA |
Period | 14/01/90 → 19/01/90 |