Photonics-based 3D radar imaging with CNN-assisted fast and noise-resistant image construction

Abstract

Photonics-based high-resolution 3D radar imaging is demonstrated in which a convolutional neural network (CNN)-assisted back projection (BP) imaging method is applied to implement fast and noise-resistant image construction. The proposed system uses a 2D radar array with each element being a broadband radar transceiver realized by microwave photonic frequency multiplication and mixing. The CNN-assisted BP image construction is achieved by mapping low-resolution images to high-resolution images with a pre-trained 3D CNN, which greatly reduces the computational complexity and enhances the imaging speed compared with basic BP image construction. Besides, using noise-free or low-noise ground truth images for training the CNN, the CNN-assisted BP imaging method can suppress the noises, which helps to generate high-quality images. In the experiment, 3D radar imaging with a K-band photonics-based radar having a bandwidth of 8 GHz is performed, in which the imaging speed is enhanced by a factor of ∼55.3 using the CNN-assisted BP imaging method. By comparing the peak signal to noise ratios (PSNR) of the generated images, the noise-resistant capability of the CNN-assisted BP method is soundly verified.

Optical color routing enabled by deep learning

“…At the deep learning level, one widely employed model is that of the convolutional neural network (CNN), a forward-propagating neural network that extracts features from input data for classification or regression. 84,85 CNNs are commonly applied to image processing tasks, utilizing convolutional layers, pooling layers, and fully connected layers to learn features inherent in input data. In contrast, the focus of inverse design methods is on optimizing objectives, making them particularly attractive for nanophotonic color routing and associated optimization problems.…”

Efficient physics-based learned reconstruction methods for real-time 3D near-field MIMO radar imaging

“…Despite this, deep learning-based reconstruction methods have not been studied much in the literature for near-field radar imaging. Most of the proposed methods are for far-field settings in SAR/ISAR or MIMO radar imaging [37,[47][48][49][50][51][52][53][54]. In the near-field and MIMO radar imaging context, there are few works for learning-based approaches [55,56], but to the best of our knowledge, there is no DNN-based reconstruction approach developed and shown to be successful for imaging 3D extended targets with random phase.…”

Analysis of disturbances in the multi-radar environment and mitigation approaches

“…It is not only widely used in the military field, but also in the civilian field and has opened up an extremely broad market. It has been successfully applied to the ground, shipborne, airborne, and spaceborne aspects since they can improve the accuracy of detecting long-range targets in all-weather conditions [1][2][3][4]. However, electronic radar systems are also limited by various factors such as bandwidth, low signal-to-noise ratio, and limited frequency tunability [5][6].…”