In recent age, a need for automatic car recognition
system has increased much. Interesting applications of the
system include but not limit to traffic monitoring system,
parking system, vehicular crime prevention and
investigation. Many researches of CLP recognition have
been published as following. Yasuharu, Masahiro and
Daisuke [1] proposed extraction and tracking of the license
plate using Hough Transform and Voted Block Matching.
Yuntao and Qian [2] proposed Markov Random Field
(MRF) and Genetic algorithm to extracting of CLP from
video. E.R. Lee, P.K. Kim and H.J. Kim [3] used Neural
Network for color extraction and a template matching to
recognize characters. In Thailand, there are many
researches proposing different methods for license plate
recognition. Thanongsak and Kosin[4] presented the FourLayer
Backpropagation Neural Network method for CLP
recognition on 70 samples of CLP, of which 80.81% were
accurately recognized. The problem still exists that dirt and
a screw on license plate may lead to lower recognition
accuracy. Ekarat and Wutipong [5] presented Thai vehicle
license plate automatic recognition using Characters Cut
and Classification method. The experimental result showed
that 70.59% from 85 samples of Thai CLP images were
correctly recognized.
This paper presents an approach to recognize off-line
entire Thai CLP using Hausdorff Distance technique for
similarity measurement. In Thailand, Sanan Srisuk [6]
presented the recognition method using Hausdorff Distance
for Thai characters in document file, resulting 100%
accuracy rate on 66,771 Thai characters images at 15% of
disturbing signals.
The next section demonstrates an overview of the
proposed system structure. Then line segmentation and character segmentation of the upper line will be described
in details. Feature extraction used in this study will also be
explained, following by our proposed recognition process
using Hausdorff Distance technique. Finally, experimental
results and future work will be addressed.