Adaptive Quantized Algorithm for Compression of Synthetic Aperture Radar (SAR) Raw Data

Synthetic Aperture Radar (SAR) data distribution is Gaussian in nature having linear as well as nonlinear region. In order to encode these data, we wish to choose the quantization step size large enough to accommodate the maximum peak-to-peak range of the signal. On the other hand we would like to make the quantization step small so as to minimize the quantization noise, which consists of Slope overload error as well as granular error. This is further compounded by nonlinear and saturated nature of the raw data or SAR image. Extremes of these data call for use of non-uniform quantizer. An alternate approach is to adapt the properties of the quantizer to the level of the input signal. In the present paper authors have studied implementation of both the input as well as output standard deviations correlated with the mean for calculating the scale factor for minimum mean square error (MSE) or maximizing the Signal to Noise ratio (SNR). The results show that none of the approach is optimum so we have combined the techniques and run an algorithm when the linearity between the mean and standard deviation breaks. For the linear part the scale factor ? is calculated to be 1.2533 while the nonlinear part begins after the input signal average value of 47.