camera was 150 HZ. During each experiment, 3000 images of each camera were captured. As shown in Fig. 34.5, in hypersonic wind tunnel the model was fixed on the fixture, a number of target points were set on it to measure the model’s deformation, and six reference points which were relative static were distributed on the fixture to rectify the camera pose. Near P6 andP7, two acceleration transducers were installed in the model, we can contrast our measurement results with that of the acceleration transducers. And P4, P5, P8, P9 were on the corner of the model, the deformation data of these points is important to analyze the critical conditions of aircraft flutter. So in this paper, we measured the deformation of P4, P5, P6, P7, P8 and P9. In hypersonic wind tunnel, the 3D vision system may move by the influence of vibration. If the world coordinate system was set on one of the cameras, the world coordinate system will move with the camera. To analyze the movement of vision system, we plot the displacement of P1 which should be still while the camera don’t move when the world coordinate system was set on the camera. As shown in Fig. 34.6, the displacement of P1 is varying during the experiment, and the maximum displacement is more than 1.5 mm. It proves that during the experiment the vision system was moving, and the measurement results when the world coordinate system was set on the camera were incredible. As the model was fixed on the fixture tightly, we can set a dynamic world coordinate system which can be defined by three points on the fixture to get precise measurement results. As shown in Fig. 34.5, we define P1 as the origin of the world coordinate system, X orientation define by P1P2, and P1, P2, P3 is on the XOY plane. Then, during the measurement the coordinates of P1 is always (0, 0, 0). In this paper, the standard deviation of distance between P1 andP2 was measured during the experiments to verify that the self-calibration method can improve the measurement accuracy of vision system in hypersonic wind tunnel. The experiment run eight times. As shown in Table 34.1, by using the self-calibration method, the standard deviation of distance between P1 andP2 was less than 0.05 mm. While without using the self-calibration method the standard deviation of distance between P1 and P2 was no less than 0.4 mm. It proves that the self-calibration method can rectify the camera pose accurately and improve the measurement accuracy in hypersonic wind tunnel. The deformation of P4, P5, P6, P7, P8 andP9 were measured (Fig. 34.7 shows the coordinates of P4 during the measurement). And we drew the spectrogram of these points in Fig. 34.8 to analyze the vibration of model. As shown in Fig. 34.8 the inherent frequency of model was 30.8 HZ, which was equal to the measurement results of the acceleration transducer. It proves that the 3D vision system can measure the deformation of model in hypersonic wind tunnel accurately. Fig. 34.6 The displacement of P1 when the world coordinate system was set on the camera Table 34.1 Standard deviation of distance between P1 and P2 (mm) Experiments 1 2 3 4 5 6 7 8 Without self-calibration 0.478 0.489 0.495 0486 0.508 0493 0.501 0.516 With self-calibration 0.034 0.044 0.029 0.017 0.023 0.028 0.013 0.020 34 A Self-Recalibrated 3D Vision System for Accurate 3D Tracking in Hypersonic Wind Tunnel 267
RkJQdWJsaXNoZXIy MTMzNzEzMQ==