Subpixel Resolution Edge Detection Techniques for Linear Sensor Arrays

Christian Mentin, Robin Priewald, Eugen Brenner

Research output: Chapter in Book/Report/Conference proceedingConference paperpeer-review

Abstract

The precise and very fast contactless measurement
of object surfaces and edges with sub-micrometer resolution plays
an important role in many applications nowadays. Optical highprecision
measuring devices are used for that purpose. The use of
narrowband point light sources for exposing an object generates a
diffraction pattern, which can be seen by a sensor. To achieve a
much higher resolution than the pixel size of the detector, that
additional information generated by this diffraction pattern can
be used. Different algorithms for estimating the projected edge
position with sub-pixel resolution using additional diffraction
pattern information were evaluated. With a modelled image
sensor pixel size of 3.8 μm, results showed that a prediction error
for the theoretical projected edge position on the sensor of less
than 200 nm could be achieved. The influence of measurement
noise, pixel size, and ADC resolution of the pixel exposure value
are discussed for the presented algorithms.
Original languageGerman
Title of host publicationProceedings of the 2016 International Conference on Broadband Communications for Next Generation Networks and Multimedia Applications
Number of pages6
Publication statusPublished - 14 Sept 2016
Event1st International Conference on Broadband Communications for Next Generation Networks and Multimedia Applications: CoBCom 2016 - Graz, Austria
Duration: 14 Sept 201616 Sept 2016
https://www.cobcom.tugraz.at/

Conference

Conference1st International Conference on Broadband Communications for Next Generation Networks and Multimedia Applications
Abbreviated titleCoBCom 2016
Country/TerritoryAustria
CityGraz
Period14/09/1616/09/16
Internet address

Keywords

  • High-resolution imaging
  • Optical diffraction
  • Algorithm design and analysis
  • Image edge detection
  • Image sensors

Cite this