Abstract:

Multiple-input multiple-output (MIMO) arrays en- able high-resolution radar imaging by synthesizing large effective apertures using spatially distributed transmit and receive anten- nas. In near-field imaging applications, the resulting multistatic geometry leads to high computational complexity. Consequently, reducing this complexity is of significant interest. To this end, most existing fast near-field MIMO imaging approaches trade reconstruction accuracy for faster runtimes by using lower- complexity approximations of the imaging operator. In this paper, we develop an efficient direct reconstruction approach that achieves the same accuracy as backprojection while prioritizing the early reconstruction of high-energy regions of the scene. Our method is based on partitioning the channels of the MIMO array to concentric apertures with equal resolution improvements.
This enables reordering of the backprojection summation for multiresolution image refinement. By storing the backprojected pixel values in a multiresolution tree, we sequentially estimate the energy distribution of the scene and decide on the region to reconstruct next. Simulations show that, compared with standard raster scanning, the proposed approach converges faster to the final backprojection image, allowing earlier recovery of the scene content.