T. Lookabaugh et al., VARIABLE-RATE VECTOR QUANTIZATION FOR SPEECH, IMAGE, AND VIDEO COMPRESSION, IEEE transactions on communications, 41(1), 1993, pp. 186-199
The performance of a vector quantizer can be improved by using a varia
ble rate code. We apply three variable rate vector quantization system
s to speech, image, and video sources and compare them to standard vec
tor quantization and noiseless variable rate coding approaches. The sy
stems range from a simple and flexible tree-based vector quantizer to
a high performance, but complex, jointly optimized vector quantizer an
d noiseless code. The systems provide significant performance improvem
ents for subband speech coding, predictive image coding, and motion co
mpensated video, but provide only marginal improvements for vector qua
ntization of linear predictive coefficients in speech and direct vecto
r quantization of images. We suggest criteria for determining when var
iable rate vector quantization may provide significant performance imp
rovement over standard approaches.