카메라 영상기반 미세진동 3차원 측정에 의한 음원 가시화 연구

Abstract

The main focus of this study is to validate the authenticity of vibration measurement using high speed camera. High speed camera with the capability of immense frame rates opens up the possibility of capturing slight pixel differences in continuous frame. Thus, the vibrations on an object could be captured in units of pixels. Through image tracking processes camera parameters and image pixel coordinates were obtained. Then the frequency analysis from the data was performed based on the Fast Fourier Transform (FFT) method. Using this technique pure sound waves of 50, 100, and 200 Hz were generated and successfully recovered. After recovering pure sound waves, experiment to recognize multi frequency sound with high speed camera and PTV (Particle Tracking Velocimetry) algorithm has been conducted. To simulate natural environment conditions simultaneous frequency and frequency sweep were generated. Simultaneous frequency was combination of 50, 62.5, 75, 87.5, and 100 Hz frequencies, and frequency sweep was progressing starting from 0 to 2000 Hz. Then the vibration caused by the frequencies were captured by high speed camera, and the measured vibration data were used to reconstruct the original data. To evaluate the authenticity of the experiment human voice were captured instead of using pure sound waves. As a result, high speed camera and PTV algorithm method showed promising results of reconstructing human voice as well. From all the conducted experiments of pure sound frequency, multi-frequency, simultaneous frequency, frequency sweep and human voice tests the integration of high speed camera and PTV algorithm showed promising results of measuring small vibrations caused by sound. Next phase of the experiment is to measure depth information using a single high speed camera and three laser pointers. Depth information is essential when measuring vibrations because it contains 3 dimensional coordinate data. Yet a single high speed camera is not sufficient to retrieve depth information from an image. Thus, lasers are used to support measuring depth of vibration. Three lasers are shot with a triangular formation on to a mirror surface. Then when the mirror is subjected to a external force, such as vibrational force, the triangular laser formation reflected off the mirror will start to deform or drift. The deformation of triangular laser points can be used to derive the depth information of subjected vibration on the mirror. The changing displacement between the three laser points can be calculated using PTV algorithm. Integrating the high speed camera images and laser point deformations can then be used to verify or reveal the vibration on the mirror. Then neural network is applied to enhance the accuracy and speed of measuring vibration. Furthermore, the application of neural network introduces the feasibility of universal vibration measuring technique.