Detection and Visibility Estimation of Surface Defects under Various Illumination Angles using Bidirectional Reflectance Distribution Function and Local Binary Pattern

Abstract

In the development of surface defect inspection systems, the surface illumination often plays a key role for the detectability of the defect. The illumination setup is currently configured manually for each type of defect which needs to be detected. This paper presents the use of a local binary pattern operator together with the bidirectional reflectance distribution function in order to detect various surface defects and estimate their visibility. This method is useful for improving the reliability of the visual surface inspection system and shortening of the time required for conducting a feasibility study. The reflectance of the sample material is acquired through a custom-built image acquisition system, which uses a robot arm in order to automatically retrieve the data under different illumination angles. By combining texture description with reflectance information, it is possible to localize defects without prior knowledge of their characteristics, whereas the defect visibility estimation requires manual ground truth marking in order to produce realistic results.

Files

Full Paper as PDF

BibTeX Citation

DOI

10.20532/ccvw.2016.0002

https://doi.org/10.20532/ccvw.2016.0002

BibTeX

@InProceedings{10.20532/ccvw.2016.0002,
  author =       {Gospodneti{\' c}, Petra and Hirschenberger, Falco},
  title =        {Detection and Visibility Estimation of Surface Defects Under
                  Various Illumination Angles Using Bidirectional
                  Distribution Function and Local Binary Pattern},
  booktitle =    {Proceedings of the Croatian Compter Vision Workshop,
                  Year 4},
  pages =        {9-14},
  year =         2016,
  editor =       {Lon{\v c}ari{\' c}, Sven and Cupec, Robert},
  address =      {Osijek},
  month =        {October},
  organization = {Center of Excellence for Computer Vision},
  publisher =    {University of Zagreb},
  abstract =     {In development of surface defect inspection systems
                  surface illumination often plays a key role in
                  defect detectability. Illumination setup is
                  currently configured manually for each type of
                  defect which should be detected. This paper presents
                  the use of a local binary pattern operator together
                  with bidirectional reflectance distribution function
                  in order to detect various surface defects and
                  evaluate their visibility. This method is useful for
                  improving visual surface inspection system
                  reliability and shortening the time required for a
                  feasibility study. Reflectance of the sample
                  material is acquired through a custom built image
                  acquisition system, which uses a robot arm in order
                  to automatically retrieve data under different
                  illumination angles. By combining texture
                  description with reflectance information, it is
                  possible to localize defects without prior knowledge
                  of their characteristics. Whereas defect visibility
                  evaluation requires manual ground truth marking in
                  order to produce realistic results.},
  doi =          {10.20532/ccvw.2016.0002},
  url =          {https://doi.org/10.20532/ccvw.2016.0002}
}