As a novel volumetric velocimetry technique, the light field particle image velocimetry (LF-PIV) based on a single light field camera benefits from its single perspective configuration and has great potential applications in the space-constraint flow measurements. However, the severe particle elongation effects and relatively heavy computational load in the volumetric reconstruction of the tracer particles limit the practical applications of the LF-PIV. To address these issues, a pre-recognition method for optimizing the weight matrix in the volumetric reconstruction of LF-PIV is proposed in this work. The crucial pre-recognition indices significantly influencing the reconstruction accuracy and efficiency are investigated using the synthetic light field images. The feasibility of the pre-recognition-based SART (PR-SART) algorithm is further experimentally validated through the reconstructions of a light source and the measurements of the horizontal laminar flow using an assembled cage-typed light field camera. The results indicated that the PR-SART algorithm is capable of mitigating the elongation effects of the reconstructed tracer particles at different depths and improving the measurement accuracy of the instantaneous velocity field. Additionally, the computational efficiency of the PR-SART algorithm is 4 times faster than the conventional SART algorithm, while the required memory is only 1/5 of the latter.

作为一种新颖的体积测速技术，基于单个光场相机的光场粒子图像测速技术（LF-PIV）受益于其单一的视角配置，在空间约束流量测量中具有巨大的潜在应用。但是，在示踪剂颗粒的体积重建中，严重的颗粒伸长效应和相对较大的计算量限制了LF-PIV的实际应用。针对这些问题，本文提出了一种在LF-PIV容积重建中优化权重矩阵的预识别方法。使用合成光场图像研究了显着影响重建精度和效率的关键预识别指标。基于预识别的SART（PR-SART）算法的可行性通过使用一个组装的笼型光场相机通过重建光源和测量水平层流来进一步通过实验验证。结果表明，PR-SART算法能够减轻重构示踪粒子在不同深度的伸长效应，提高瞬时速度场的测量精度。此外，PR-SART算法的计算效率比常规SART算法快4倍，而所需的内存仅为后者的1/5。结果表明，PR-SART算法能够减轻重构示踪粒子在不同深度的伸长效应，提高瞬时速度场的测量精度。此外，PR-SART算法的计算效率比常规SART算法快4倍，而所需的内存仅为后者的1/5。结果表明，PR-SART算法能够减轻重构示踪粒子在不同深度的伸长效应，提高瞬时速度场的测量精度。此外，PR-SART算法的计算效率比常规SART算法快4倍，而所需的内存仅为后者的1/5。