Enhanced Microwave Imaging by Bilinear Compressed Sensing

Microwave imaging (migration) and spectral analysis are widely used in the remote sensing techniques. Generally, a known signal will be transmitted and the transfer function of the observed zone will then be recovered by the received signal and the transmit signal. Nevertheless, those scenes are usually available for homogeneous setting, where the transmit signal will not be modified by e. g. the medium and the wave propagation mechanism. In inhomogeneous case, however, the transmit signal is modified by the channel. Thus, the received signal can be formulated as a bilinear function over transmit signal and channel transfer function. Additionally, the measured data in practice has relative low dimension property due to some technique as well as physical limitations. Regarding e. g. a super-resolution problem, its objective vectors exist usually in the higher dimensional space. Therefore, the corresponding inverse problem can be very ill-posed. Compressed Sensing (CS), which is popular for dealing with such ill-conditioned cases, will be introduced. Particularly, the question of how to realize the bilinear CS model in our special case will be discussed. Both theory and practical tests show its promising performance.