Improved Ray Tracing Based Augmentations for Concealed Object Detection in Activate Millimeter Wave Imaging

Active millimeter wave (AMMW) imaging technique plays an important role in the security check, due to its ability to detect the concealed objects harmlessly without violating privacy. With the rapid development of deep learning, several algorithms have achieved great success in concealed object detection for AMMW images. However, their performance is restricted for the lack of sufficient data, which is the prerequisite for training the deep learning neural network. To address this task, an improved ray tracing (RT) method is proposed to fast solve scattering fields. Specifically, the Blinn-Phong model, which is a simple but effective model in computer graphics rendering, is utilized to simulate the scattering in the target surface. Comparing with the conventional full-wave simulation methods, the proposed method can achieve similar results with much less time. Then the range migration algorithm (RMA) is employed to reconstruct the images. Both visual and numerical results, as well as comparisons with other classical simulation methods, demonstrate the effectiveness and superiority of the proposed improved ray tracing framework in generating AMMW images for deep learning based concealed object detection.