The weld inspection system developed in this work is based on the principle of optical profiling. Optical profiling is a method for visualising the surface profile of an object while it is being viewed by an imaging source which is perpendicular to its surface. The signal to be viewed by the imaging source is provided by a line of laser light. The laser light is angled relative to the weld surface it is incident upon. The distortion in the laser line gives an indication of weld surface quality. A traditional imaging source for such systems has been the CCD camera. This has been replaced with a novel CMOS camera. The two main benefits a CMOS camera has over standard CCD cameras are (1) Random Pixel Access (2) Logarithmic Response to Light. Taking advantage of the random pixel access capabilities of the CMOS camera procedures were developed to extract the information relevant to weld analysis from minimal sections of the full 2-D image. Using the grey scale signals (logarithmic scale) from just a few single rows or lines within the 2-D image of the projected laser line, it was possible to (a) determine weld edges using wavelet analysis (b) achieve weld surface classification using multivariate analysis. These procedures have been implemented in laboratory set ups, but were also tested successfully online during laser welding.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:637003 |
| Date | January 2002 |
| Creators | Francis, R. |
| Publisher | Swansea University |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
Page generated in 0.002 seconds