Analysis of Surface Images with Matlab

-- MicheleLanzetta? - 12 Aug 2004

  • Matlab code (see below)
  • Samples of images grabbed using the colour Logitech QuickCam Pro 4000 (Images.zip): Concrete, coin and PCB
Nominal resolution is 1280 x 960 @ 24 bit.


  • Difference of two images
    CoinDiff.jpg
One with (lab) environmental light and one with directional 100 W halogen light at low incidence angle. A dime is shown, so the high magnification is clear. The relative coin-camera displacement/vibration between the two acquisitions is visible (out-of-focus like effect). It can also be noticed that due to light-camera-object configuration, the bottom part of the image (further apart from light source) shows more asperities than the top part of the image (closer to the light source) as a consequence of the change in the angle of incidence of light. The radial optical distortion (cushion effect) is also shown.

This image processing m-file (impr1.m) reproduces the output shown below and detailed above.

  • R G B colour bands subtraction:
    CoinDiffx3.jpg

The different response of the metal surface to the three light radiation components is shown. This suggests using a colour light source and just perform a difference of images between two of the three components, to reduce the acquisition time and to avoid the displacement problem described above.

Now I am designing a led based colour lighting device to have different light components from different directions. Any idea is welcome.

 
This site is powered by the TWiki collaboration platformCopyright &© by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback