A fast, non-iterative ray-intersection approach for three-dimensional microscale particle tracking

We propose a non-iterative ray tracing method with robust post-capture microlens array sensor alignment to reconstruct sparse particle concentration in light field particle image velocimetry and particle tracking velocimetry nearly instantaneously. Voxels traversed by various rays are stored by a kd-tree to reduce memory load and computational time. A cloud point classification algorithm is employed for particle identification and spatial reconstruction. The approach is tested with a physically-based realistic model of a light field camera. Also, an optical system is assembled in a microscope to directly obtain the 3D laminar velocity field in the fully-developed region, which exhibits good agreement with the theoretical solution.

Automated summary

We propose a non-iterative ray tracing method with robust post-capture microlens array sensor alignment to reconstruct sparse particle concentration in light field particle image velocimetry and particle tracking velocimetry nearly instantaneously. The method utilizes kd-tree for storing voxels traversed by rays to reduce memory load and computational time, and employs a cloud point classification algorithm for particle identification and spatial reconstruction. Experimental results demonstrate the effectiveness of the proposed method, and its application in a microscope shows good agreement with theoretical solution.