Vision-based vibration measurement of machine tool

Abstract

This study aims to develop a visionbased measurement method for machine tool vibrations and dynamic motion errors. In this study, the potential of the visionbased method in high resolution measurement is evaluated. A model fitting method is considered for the image processing that is appropriate for mechanical vibration measurement. The influence of lighting on the measurement resolution is also evaluated. In the method, the displacement of the marker attached to the object is obtained through a video captured by using a highspeed camera. To obtain the marker displacement, the center position of the marker is tracked by fitting a quadratic function to the intensity distribution of the marker. Experiments are conducted to evaluate the measurement resolution of the model fitting method. The vibration displacement of a cantilever with various amplitudes was measured using the vision based method under different lighting conditions. The measured displacement is analyzed in the frequency domain to determine the measurement resolution and noise level. In addition to the model fitting method, an edge detection method and a threshold detection method are used for the marker position detection and compared with the model fitting method. The experiment results indicate that the model fitting method has the ability to detect the minimum amplitude of 1/53 pixels that corresponds to a 3.8 µm resolution in 120 × 160 mm range.