Bit Allocation and Transform Coding

Abstract

Often a signal coding system contains several different quantizers, each of which has the task of encoding a different parameter that is needed to characterize the signal. Each such parameter may have a different physical meaning and may require a different relative accuracy of reproduction to achieve a desired overall quality of the reconstructed signal. The number of bits available to collectively describe this set of parameters is inevitably limited. Consequently, a major concern of the designer of a coding system is bit allocation, the task of distributing a given quota of bits to the various quantizers to optimize the overall coder performance. In a broader sense, this is an interdisciplinary problem which arises in everyday life: how do we efficiently distribute a limited or scarce resource to achieve the maximum benefit?