Publication record · 18.cifr/1982.lloyd.quantization-pcm
18.cifr/1982.lloyd.quantization-pcmIt has long been realized that in pulse-code modulation (PCM), with a given ensemble of signals to handle, the quantum values should be spaced more closely in the voltage regions where the signal amplitude is more likely to fall. It has been shown by Panter and Dite that, in the limit of a large number of quanta, the quantization noise power varies inversely as the cube of the number of quanta per unit of amplitude range. In this paper, necessary conditions for the optimum choice of both quantum values and partition values are derived. These conditions lead to an iterative procedure for finding the optimum solution.
Computing related research...
Loading DOI…
Sign in to run agents. GPU access requires an institutional membership.
How to get GPU access: Your university, lab, or company can become a CIFR institutional member. Members get GPU-accelerated runs for all their researchers. Contact us
No invocations yet — be the first to call this agent.
The algorithm converges to local optima only; global optimization and initialization strategies remain open. Direct extension to vector quantization (LBG algorithm) and adaptive quantization for non-stationary sources are natural follow-ups. Tight rate-distortion bounds for specific distribution families were also identified as open problems.