Prof. Peter Lehmann
Universität Kassel, Germany
Title: Resilient high-resolution surface topography inspection techniques
for machine vision systems
Abstract:
Machine vision systems typically acquire two-dimensional (2D) images in order to obtain relevant surface texture features of an object of interest, and even if AI methods may help to extract 3D information from 2D images, optical 3D imaging techniques are still seen as more trustworthy. However, in machining environments conventional 3D imaging techniques such as fringe projection, focus variation, or confocal microscopy fail due to motion blurring and environmental vibrations. These difficulties get worse, if methods of highest sensitivity such as phase shifting or coherence scanning interferometry come into play.
This contribution shows, how well-established 3D measurement techniques can be modified in order to fulfill the requirements of machine vision. We show exemplary results of current research projects dealing with in-situ measurement of 3D surface topography.
One option to obtain 3D information from a single 2D image is based on single-shot fringe pattern profilometry (FPP). We demonstrate that FPP can be used to obtain 3D information of surface texture and damages of civil infrastructure devices such as wind turbines. The pulsed projection of a fixed Ronchi grating in combination with spatially resolved phase analysis of the fringe pattern captured by the camera enables the detection of small surface delamination and spalling defects on the leading edge of rotor blades of wind turbines. The phase analysis is based on a discrete short-time Fourier transform along the fringes of the captured image. We developed this technique in order to perform a drone-based in-situ inspection of rotor blades.
The second example demonstrates how one can adapt coherence scanning interferometry (CSI) in harsh machining environments. CSI requires a so-called depth scan, where an interferometer moves towards the measurement object in order to obtain an interferometric image stack. This is needed in order to reconstruct the surface topography requiring equidistant sampling of interferograms and appropriate signal processing algorithms. Due to the depth can CSI instruments strongly suffer from environmental vibrations. We integrated an interferometric laser optic distance sensor (IDS) into custom-made CSI systems, that records distance changes between the object and the interferometer with nanometer resolution and high data rates of several tens of kHz. Since the image acquisition by the camera is synchronized with the IDS distance measurement, the IDS data allow us to correct for vibration disturbances by rearranging the measured image stack and thus compensating for the vibration influences.
Biodata: TBA