Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Digital image processing for the automation of NDT by means of endoscopy

 
: Spinnler, Klaus; Bergen, Tobias; Sandvoss, Jan; Wittenberg, Thomas

Purschke, M. ; International Committee for Non-Destructive Testing -ICNDT-; Deutsche Gesellschaft für Zerstörungsfreie Prüfung e.V. -DGZfP-, Berlin:
19th World Conference on Non-Destructive Testing, WCNDT 2016 : Munich, Gemany, 13-17 June 2016; Proceedings; USB-Stick
Berlin: DGZfP, 2016 (DGZfP-Berichtsbände 158)
ISBN: 978-3-940283-78-8
S.1168-1175
World Conference on Non-Destructive Testing (WCNDT) <19, 2016, Munich>
Englisch
Konferenzbeitrag
Fraunhofer IIS ()
Visualisierung; industrielle BV; Endoskope; Defekterkennung; Bewegungserfassung; Algorithmen; 3D Bildverarbeitung

Abstract
Visual inspection is one of the most widely used NDT-methods. Endoscopes are an important tool for the inspection of difficult to access surfaces inside the work pieces under test. The application of video technology enables the ergonomic visualization of the inspection. The application of digital technology enables even more: Digital image processing can contribute to an enhanced visualization. From endoscopic image sequences, so called panoramic images can be computed, which show an overall view of a cavity and allow for the check of an inspection result at a glance. By means of filtering and masking of the recorded images, poorly exposed image portions, e. g. underexposed or overexposed, can be discarded. Thus overall views with nearly homogeneous illumination level can be presented to the user. Fiberscopes show images with strong artefacts due to the fiber-optic image transmission. The superimposed honeycomb structure observed in these images can be eliminated by digital image processing and thus an enhanced image can be presented to the user. Once an enhanced image, prepared in the described way, is achieved, any method for automatic defect recognition by means of image processing can be applied. These comprise in general the processing steps segmentation, feature extraction, classification, up to the final rating of the defect situation or the surface quality of a workpiece. In conjunction with data of the movement of the endoscope, also a 3D reconstruction of the cavity under inspection is possible. Such algorithms are based on the principle of “shape from-motion” or “shape-from-polarisation” for example. Typical applications are automated inspections of hydraulic components with bore diameters in the range of 2 – 20 mm in automotive industry. By means of an automated materials handling and positioning technology a fully automated surface inspection of bore holes and cavities in workpieces made from metal or plastics can be done. Thus a 100-percent, process-integrated routine test with high quantities and short cycle times is possible.

: http://publica.fraunhofer.de/dokumente/N-473123.html