面向瞬态成像仿真的遮挡目标逆向重构方法

由于瞬态成像硬件系统存在结构复杂、价格昂贵以及性能评估困难等问题，提出了一种新的逆向重构算法，并利用软件仿真的方法建立了瞬态成像链路．该方法首先根据仿真场景简化像素累积函数，并引入检测概率的方法获得三维空间中目标体素的采样; 然后利用射线追踪的方法计算目标的全局检测概率，通过阈值分割得到目标的三维点云信息;最后融合点云进行三维场景渲染．实验结果表明，这种方法能够仿真遮挡目标三维场景的瞬态成像过程，并且逆向重构算法增强了遮挡目标重构的效果．

Reconstruction of the hidden 3D shape for transient imaging simulation

With the development of ultrafast imaging devices, transient imaging can record the movement of light on the femtosecond scale. One of its applications is the reconstruction of hidden 3D shapes, which is to reconstruct the three-dimensional structure of objects hidden outside the camera field. Due to the complex structure, the high price and difficult performance-evaluation of the transient imaging hardware system, this paper presents a new reverse reconstruction algorithm and establishes a transient imaging process using software simulation. Beginning with a simplified heatmap function according to the scene, this reverse algorithm introduces the detecting probability to obtain the sampled data of the object in the scene; after that, it computes the global detecting probability of the object using ray-tracing methods and gets a 3D point cloud by thresholding results; finally, the point cloud is visualized in the 3D space. The result shows that the simulation method presented in this paper is able to describe the transient imaging process of a hidden object in a 3D scene, and that the reverse reconstruction algorithm can improve the reconstruction of hidden objects.