A MATLAB SMO Implementation to Train a SVM Classifier: Application to Multi-Style License Plate Numbers Recognition
Pablo Negri
→ BibTeX
@article{ipol.2018.173,
    title   = {{A MATLAB SMO Implementation to Train a SVM Classifier: Application to Multi-Style License Plate Numbers Recognition}},
    author  = {Negri, Pablo},
    journal = {{Image Processing On Line}},
    volume  = {8},
    pages   = {51--70},
    year    = {2018},
    doi     = {10.5201/ipol.2018.173},
}
% if your bibliography style doesn't support doi fields:
    note    = {\url{https://doi.org/10.5201/ipol.2018.173}}
published
2018-05-22
reference
Pablo Negri, A MATLAB SMO Implementation to Train a SVM Classifier: Application to Multi-Style License Plate Numbers Recognition, Image Processing On Line, 8 (2018), pp. 51–70. https://doi.org/10.5201/ipol.2018.173

Communicated by Martín Rais
Demo edited by Martín Rais

Abstract

This paper implements the Support Vector Machine (SVM) training procedure proposed by John Platt denominated Sequential Minimimal Optimization (SMO). The application of this system involves a multi-style license plate characters recognition identifying numbers from '0' to '9'. In order to be robust against license plates with different character/background colors, the characters (numbers) visual information is encoded using Histograms of Oriented Gradients (HOG). A reliability measure to validate the system outputs is also proposed. Several tests are performed to evaluate the sensitivity of the algorithm to different parameters and kernel functions.

Download

Supplementary Materials

Dataset of license plate numbers from four countries having different fonts, and character/background colors: dataset