Inverse synthetic aperture radar imaging method combining gate unit and transfer learning

Abstract

The invention provides an inverse synthetic aperture radar imaging method combining a gate unit and transfer learning, and the method comprises the steps that a full convolutional neural network FCNN is constructed through employing a conventional convolution layer, an activation function layer, a batch normalization layer, a maximization pooling layer, and the like; a plurality of paths are established by utilizing jump connection SKs in the FCNN and are used for directly transmitting network shallow feature representation OFRs to a network reconstruction layer; secondly, the gate unit is further introduced into the FCNN to form a G-FCNN, and a transfer learning strategy TL is adopted to ensure that the performance of the G-FCNN is optimal; and then large-scale radar training data is constructed by using electromagnetic simulation software for pre-training the G-FCNN, and network layer parameters in the pre-trained G-FCNN are finely adjusted by using small-scale actually measured radar data to obtain optimal network parameters for a target imaging task. The inverse synthetic aperture radar imaging method is superior to an existing ISAR imaging method based on the convolutional neural network.

Description

technical field [0001] The invention belongs to the technical field of radar signal processing, and relates to a method for sparse ISAR imaging. Background technique [0002] ISAR (Inverse synthetic aperture radar) can obtain high-resolution images of moving targets under all-weather and all-time conditions, and is an important tool for non-cooperative target surveillance and identification. The traditional RD (Range doppler, range Doppler) method uses FFT to realize target azimuth imaging. The RD imaging method has high imaging efficiency, but its imaging results are susceptible to sidelobe interference. The sparse ISAR imaging method can reconstruct the target image with less sidelobe interference and high image contrast by using very few measurement values. However, the imaging quality and efficiency of sparse ISAR imaging methods are limited by the inaccurate sparse representation and the time-consuming iterative reconstruction process, respectively. [0003] Learned ...

AI Technical Summary

Problems solved by technology

However, the OFRs delivered by SKs inevitably contain the feature information of false scatter points, which makes it easy to appear false scatter points in the final reconstructed target image.

Images