3D Scattering Tomography by Deep Learning with Architecture Tailored to Cloud Fields
This work addresses the problem of computationally expensive 3D reconstruction of atmospheric cloud fields, which is crucial for meteorological modeling and climate science.
This paper introduces 3DeepCT, a deep neural network designed for 3D reconstruction of scattering volumes, specifically atmospheric cloud fields, from multi-view images. It significantly outperforms physics-based inverse scattering methods in both accuracy and computational time, offering orders of magnitude improvement.
We present 3DeepCT, a deep neural network for computed tomography, which performs 3D reconstruction of scattering volumes from multi-view images. Our architecture is dictated by the stationary nature of atmospheric cloud fields. The task of volumetric scattering tomography aims at recovering a volume from its 2D projections. This problem has been studied extensively, leading, to diverse inverse methods based on signal processing and physics models. However, such techniques are typically iterative, exhibiting high computational load and long convergence time. We show that 3DeepCT outperforms physics-based inverse scattering methods in term of accuracy as well as offering a significant orders of magnitude improvement in computational time. To further improve the recovery accuracy, we introduce a hybrid model that combines 3DeepCT and physics-based method. The resultant hybrid technique enjoys fast inference time and improved recovery performance.