Optimum data quantization in microwave imagery and its effect upon image quality

An optimum quantization scheme for compressing the aperture data in a microwave antenna array is derived. The optimization criterion is maximization of the quality of the output image from the system. The procedure can be applied to a quantizer designed for any number of bits per sample. It is shown that hard limiting the input to a phased array while leaving the phase unaltered preserves much of the image integrity. Amplitude-only information destroys it. Hard limiting plus one bit of phase information sometimes suffices. Hard limiting plus two bits of signal phase per sample often results in acceptable imagery. A minimum of 4 bits of information per complex signal sample guarantees a relatively good image, provided that 3 bits are used to represent the phase and 1 bit is used to represent the amplitude. Eight bits are required to assure a high-fidelity image. When the number of bits per data sample is no more than 8, the number of bits assigned to phase should be two larger than the number of bits assigned to amplitude. For finer quantization, equal numbers of bits should be assigned to amplitude and phase.<
>