Imaging Algorithm and Measurement Error Impact on Retrievals from the Microwave Single Pixel Imager (MSPI)

The performance of the imaging algorithm for a single pixel imaging microwave radiometer is discussed in terms of the pixel-by-pixel errors in the retrieved images. There are two limiting factors - the achievable performance of the compressive sensing imaging algorithm itself, and the impact of noisy measurements that feed the algorithm. The limits of the currently used algorithm, in the absence of measurement noise, are provided for four different simulated scenes. Significant improvement is possible through inclusion of a priori information, and is quantified. The key terms for the error budget are discussed and are used to determine an expected value for the noise on measurements with the current configuration. The imaging errors in the presence of this noise are given. Averaging multiple images can reduce the error, and values for averaged images are also given.