Non-line-of-sight (NLOS) imaging technology is to ‘see’ the target out of sight, such as an object around a corner or hidden by some shelters. However, due to constraints of device definition and computing load, NLOS system is usually expensive and requires hidden objects with special material and simple shape. Besides, imaging space of system is limited. We perform a series of simulation with 1550nm infrared laser to expand the application field and improve the performance of NLOS system. Based on math and physical properties, main experimental components are modeled and data acquisition process is completed first. Then, the ellipsoid inversion algorithm is used to reconstruct the hidden space and imaging results are obtained. Finally, multiple series of system parameters are set and their influence on imaging results is analyzed. Results demonstrate that echo signal intensity after multiple reflections provides adequate information to reconstruct the geometry of a hidden object. However, the number of laser scanning position, resolution of detector, voxel division and location of the scanning area will all have a critical influence on NLOS imaging results.
Non-line-of-sight (NLOS) imaging is an emerging technique, which can observe objects obscured by occluders. Thanks to the improvement of optical configurations, it is receiving growing interest from researchers. In this paper, we reconstruct both 2D and 3D images by adopting the light-cone transform and validated on simulated data. Numerical results are evaluated by structural similarity index (SSIM). The results showed the good performance of the algorithm in preserving the details of 2D image and reconstruction of 3D image. The structural similarity index of the reconstructed image and the reference image is more than 50%, the target is hence being identified. This work contributes to the construction of the real system.
This passage studies on theory and scene simulation of NLOS imaging. Based on math and physical properties, a simulation platform is built for a NLOS imaging system, including a femtosecond laser, a scanning galvo system, a lambertian surface, several hidden scenes, an ultrafast photodetector to transfer the intensity of laser echo signal to voltage value and a TCSPC module to produce intensity-time histograms. By the simulation platform on MATLAB, precise imaging of the scenes is accomplished. Results show that multi-path analysis using echo signal intensity versus time provides enough proof to reconstruct 3D geometry of a hidden scene.