Experimental Evaluation of Vehicle Detection Based on Background Modelling in Daytime and Night-Time Video

Abstract

Vision-based detection of vehicles at urban intersections is an interesting alternative to commonly applied hardware solutions such as inductive loops. The standard approach to that problem is based on a background model consisting of independent per-pixel Gaussian mixtures. However, there are several notable shortcomings of that approach, including large computational complexity, blending of stopped vehicles with background and sensitivity to changes in image acquisition parameters (gain, exposure). We address these problems by proposing the following three improvements: (i) dispersed and delayed background modelling, (ii) modelling patch gradient distributions instead of absolute values of individual pixels, and (iii) significant speed-up through use of integral images. We present a detailed performance comparison on a realistic dataset with handcrafted groundtruth information. The obtained results indicate that significant gains with respect to the standard approach can be obtained both in performance and computational speed. Experiments suggest that the proposed combined technique would enable robust real-time performance on a lowcost embedded computer.

Files

Full Paper as PDF

BibTeX Citation

DOI

10.20532/ccvw.2014.0014

https://doi.org/10.20532/ccvw.2014.0014

BibTeX

@InProceedings{10.20532/ccvw.2014.0014,
  author =       {Igor Lipovac and Tomislav Hrka{\' c} and Karla
                  Brki{\' c} and Zoran Kalafati{\' c} and Sini{\v s}a
                  {\v S}egvi{\' c}},
  title =        {Experimental Evaluation of Vehicle Detection based
                  on Background Modelling in Daytime and Night-Time
                  Video},
  booktitle =    {Proceedings of the Croatian Compter Vision Workshop,
                  Year 2},
  pages =        {3-8},
  year =         2014,
  editor =       {Lon{\v c}ari{\' c}, Sven and Suba{\v s}i{\' c},
                  Marko},
  address =      {Zagreb},
  month =        {September},
  organization = {Center of Excellence for Computer Vision},
  publisher =    {University of Zagreb},
  abstract =     {Vision-based detection of vehicles at urban
                  intersections is an interesting alternative to
                  commonly applied hardware solutions such as
                  inductive loops. The standard approach to that
                  problem is based on a background model consisting of
                  independent per-pixel Gaussian mixtures. However,
                  there are several notable shortcomings of that
                  approach, including large computational complexity,
                  blending of stopped vehicles with background and
                  sensitivity to changes in image acquisition
                  parameters (gain, exposure). We address these
                  problems by proposing the following three
                  improvements: (i) dispersed and delayed background
                  modeling, (ii) modeling patch gradient distributions
                  instead of absolute values of individual pixels, and
                  (iii) significant speed-up through use of integral
                  images. We present a detailed performance comparison
                  on a realistic dataset with handcrafted groundtruth
                  information. The obtained results indicate that
                  significant gains with respect to the standard
                  approach can be obtained both in performance and
                  computational speed. Experiments suggest that the
                  proposed combined technique would enable robust
                  real-time performance on a low-cost embedded
                  computer.},
  doi =          {10.20532/ccvw.2014.0014},
  url =          {https://doi.org/10.20532/ccvw.2014.0014}
}