Sparse decoupling imaging network for wideband two‐dimensional MIMO radar imaging using model‐assisted and attention mechanism

The problem of sparse decoupling radar imaging methods based on deep learning is researched. An improved model‐driven learning imaging network with a complex‐valued convolution block attention module plugged into each sub‐network is proposed. This method can solve the high sidelobe and coupling problem in sparse wideband Multiple‐Input Multiple‐Output (MIMO) radar. In addition, it can better focus on the target area and capture target information to boost model representation power. Experimental results verify the validity of the proposed method.