Seeing Sound – A New Method For Measuring Vibrations With Cameras

Vibrations can be recorded in videos as periodic fluctuations in the brightness of individual pixels and quantified using Fourier transforms. Possible applications include the localisation of sound bodies or the investigation of industrial plants for unwanted vibrations. However, as the Nyquist-Shannon sampling theorem states, the highest frequency that can be clearly identified depends on the frame rate of the camera used, which is why a high-speed camera is usually required. This paper presents a concept in which high frequencies can also be correctly identified with a smartphone by taking measurements at irregular time intervals. Calibration methods were developed and an algorithm for analysing the video material was implemented and optimised in Python.