Large-scale particle image velocimetry (LSPIV) has been increasingly applied in whole-field flow measurement for its non-contact operation mode. However, due to the complicated flow conditions and optical disturbances on the water-air interface, commonly-used imaging technologies are degenerated by the low visibility and image distortion. There are also increased challenges for control point calibration. This paper proposes an imaging measurement method that imitates the imaging computation mechanism in compound eyes. The method is independent of ground control points and can achieve flow measurement with natural patterns, which can solve the problem caused by the lack of visible tracers. Field evaluation results demonstrate that this is the first method to achieve real-world whole-field flow measurement in challenging conditions. Specifically, the velocimetry error is less than 5%, while ranging from 5% to 15% for other LSPIV systems. The method meets current hydrological standards and acquires flow data in wild rivers, performance that is realized with difficulty by other methods.

大规模粒子图像测速技术（LSPIV）因其非接触式操作模式而越来越多地应用于全场流量测量中。然而，由于复杂的流动条件和水-空气界面上的光学干扰，低能见度和图像失真导致了常用的成像技术的退化。控制点校准也面临着越来越多的挑战。本文提出了一种模仿双眼成像计算机制的成像测量方法。该方法与地面控制点无关，可以实现自然模式的流量测量，可以解决由于缺乏可见示踪剂而引起的问题。现场评估结果表明，这是在挑战性条件下实现真实世界全场流量测量的第一种方法。具体来说，测速误差小于5％，而其他LSPIV系统的测速误差为5％至15％。该方法符合目前的水文标准，并能采集野河中的流量数据，而其他方法难以实现其性能。