Automated Computer Vision-Based Reading of Residential Meters

Abstract

We present a solution for automated reading of various residential meters from photographs, using computer vision algorithm. The solution has several modules that are executed sequentially: meter type recognition, geometric transform of images, ROI extraction and OCR. Algorithm uses SURF and HOG features to detect and describe feature points used for device recognition and OCR. The final goal is to provide complete counter values and complete serial number of the meter. The solution allows for a limited amounts of poor imaging conditions, and usually fails in poor image conditions when even human observers would have difficulties reading meters. The solution has been implemented in MATLAB environment and its computer vision library. An initial image database has been collected for testing purposes. Test results are reported in the paper.

Files

Full Paper as PDF

BibTeX Citation

DOI

10.20532/ccvw.2016.0015

https://doi.org/10.20532/ccvw.2016.0015

BibTeX

@InProceedings{10.20532/ccvw.2016.0015,
  author =       {Ko{\v s}{\v c}evi{\' c}, Karlo and Suba{\v s}i{\'
                  c}, Marko},
  title =        {Automated Computer Vision-based Reading of
                  Residential Meters},
  booktitle =    {Proceedings of the Croatian Compter Vision Workshop,
                  Year 4},
  pages =        {24-29},
  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 =     {We present a solution for automated reading of
                  various residential meters from photographs, using
                  computer vision algorithm. The solution has several
                  modules that are executed sequentially: meter type
                  recognition, geometric transform of images, ROI
                  extraction and OCR. Algorithm uses SURF and HOG
                  features to detect and describe feature points used
                  for device recognition and OCR. The final goal is to
                  provide complete counter values and complete serial
                  number of the meter. The solution allows for a
                  limited amounts of poor imaging conditions, and
                  usually fails in poor image conditions when even
                  human observers would have difficulties reading
                  meters. The solution has been implemented in MATLAB
                  environment and its computer vision library. An
                  initial image database has been collected for
                  testing purposes. Test results are reported in the
                  paper.},
  doi =          {10.20532/ccvw.2016.0015},
  url =          {https://doi.org/10.20532/ccvw.2016.0015}
}