Integration of mmWave FMCW Radar and Stereo Camera for 3D automotive SAR Imaging

The millimeter-wave (mmWave) Synthetic Aperture Radar (SAR) systems are widely studied in the automotive industry. Traditional 2D mmWave SAR systems, however, have faced challenges in accurately recognizing the shapes of objects and in providing height information. This study introduced a novel 3D automotive mmWave SAR system that integrates a 77 GHz Frequency Modulation Continuous Wave (FMCW) radar with a stereo camera. This system processes radar signals at each point of the point cloud generated from the stereo camera to reconstruct well-focused 3D SAR images with height information. By using an entropy-based gradient descent method for the optimized six Degrees of Freedom (DOF) extraction, the research achieved radar and stereo camera alignment, significantly enhancing phase estimation accuracy and target positioning. Furthermore, radiometric calibrations–both relative and absolute–were performed, achieving Relative Calibration Accuracy (RCA) of 0.13 dB and Absolute Calibration Accuracy (ACA) of 0.15 dB. For the validation of our system, the 3D SAR images of the street tree were reconstructed, demonstrating enhanced capability in accurately depicting the shapes of the complex object with backscattering coefficients. The comparative analysis demonstrated significant improvements over traditional methods, with Mean Square Error (MSE) reduced to 0.42, Peak Signal to Noise Ratio (PSNR) increased to 6.29, and Structural Similarity Index (SSIM) increased to 0.35. This SAR system significantly enhances the potential for innovative automotive applications by providing accurate object information.