Non-line-of-sight target 2D size detection with few channels at a time

Non -line -of -sight (NLOS) imaging techniques have garnered significant interest in recent years for their ability to reconstruct obscured scenes. This type of imaging typically needs numerous temporal histograms, which require thousands of scans on the relay surface. Reconstructing obscured targets using several channels, i.e., capturing echoes from a few positions of the relay surface, is an attractive prospect but has so far been restricted to localization applications. Herein, we propose a rectangle -based data -driven method and first demonstrate that the 2D size of the hidden target can be detected even with as few as 2 or 3 detectors by exploiting the information inside an echo signal. Specifically, we conceptualize the occluded target as a rectangle representing the target size and train a multi -layer perceptron to learn the correspondence of the hidden scene from the generated data, in which the method can directly respond to unknown targets. Both numerical simulations and experiments are performed to evaluate the feasibility of this method. Furthermore, this work provides a feasible way of detecting hidden scene without scanning.