Two-Dimensional Array Processing with Compressed Sensing

Front-view radar images of large moving targets such as humans are difficult to capture with existing radar techniques such as SAR and two-dimensional array processing. SAR images may be significantly distorted by the motion of the human. Array processing may require a large number of sensors which may make the radar costly and complex. In this paper, we propose to reduce the number of sensors required for generating frontal images by combining compressed sensing with array processing. First, we apply a compressed sensing based reconstruction technique to generate frontal images of a moving phantom target on a frame to frame basis. In the second technique, we exploit the temporal correlation across multiple frames using Kronecker compressed sensing to generate the radar image of each frame. Both approaches are implemented with 25% and 50% of the number of sensors that would be required by conventional two-dimensional array processing.