Millimeter Wave Radar Image Aliasing Enhancement Method Based on CycleGAN

The millimeter wave imaging platform designed based on the principle of MIMO-SAR technology made up for the shortage of a large number of sensors required by the array imaging platforms of synthetic aperture technology, and achieved the same or even better effects using a single radar scan. However, the long scanning has limited practical applications of the platform. Increasing the scanning interval can save a lot of scanning time, but it also brings the problem of undersampling resulting in the aliasing of objects in the images. Superresolution technology can not solve the overlapping phenomenon, so it is necessary to find an effective method to deal with the overlapping phenomenon for image enhancement. In this paper, we proposed a CycleGAN-based millimeter wave radar image aliasing enhancement method to transform the aliased millimeter-wave images into clear images, thus significantly reducing the scanning time required by the imaging platform of array millimeter wave radar under the condition of ensuring the imaging effect. The experimental methods were evaluated using peak signal-to-noise ratio (PSNR) and structural similarity (SSIM). The average PSNR and SSIM of this method are 11.37 dB and 0.46, respectively, which effectively improves the image enhancement effect of the under-sampling problem caused by the large sampling interval.