Perceptually optimum gray-scale transformation of multibin time history data

Gray-scale displays of a time history of data in multiple channels are encountered in various applications. The signal tracks in a noisy display often have to be detected by a human operator. To exploit the pattern-recognition capability of the human "receiver", the perceptual factors influencing visual discrimination needs to be incorporated into the gray-scale transformation scheme of the data. To this end, an objective function, viz. the total perceived entropy, is proposed in this paper for optimum gray-scale quantizer design. Two types of quantizers, one designed with unconstrained maximization of the objective function and another which maximizes the objective function with the constraint enforced by Weber's law, are proposed. It is demonstrated, using simulated spectrogram data, that these quantizers provide better signal track discrimination than Lloyd-Max quantization.