One-bit SAR Imaging Based on Perceptron Learning Algorithm with Bootstrap Method

The one-bit quantization with time-varying thresholds has been studied in the field of compressed sensing and SAR imaging. The Perceptron Learning Algorithm (PLA) uses a sign function as its activation function, and this is consistent with the one-bit quantization model. Therefore, one-bit SAR imaging based on PLA shows better imaging performance than the existing approach based on Logistic Regression Algorithm (LRA). Moreover, the bootstrap method is introduced into PLA to improve the imaging performance at low SNR. Experimental results validate the effectiveness of the proposed method.