Abstract
Reconstructing an object’s location and monitoring its movement through scattering media remains a significant challenge in applications. Existing methods suffer from the object motion limit, the prior of the object information, or the complex optical setup. Here, we focus on reconstructing the 3D location and trajectory of an object in motion behind scattering media by explicitly modeling and inverting the time-varying light transportation. A time-varying scattering imaging model is derived to encode the positions of the moving object in the intensity videos captured by a conventional RGB camera. Based on the model, we find that the object lies on 3D surfaces determined by point pairs on the scattering media. We then develop a back-projection method to build a 3D confidence map for the voxelized object space to find the voxel with the maximum confidence as the object position in the reconstructed trajectory at the corresponding video time. The effectiveness of the proposed method to locate moving self-illuminated and light-reflective objects in different shapes behind scattering media with different thicknesses using 2D intensity images is verified by simulated experiments and real scattering imaging systems. The reconstructions of multiple objects and different lighting conditions are discussed.
