@article{10.11648/j.mcs.20210606.13,
  author = {Gulsum Cigdem Cavdaroglu and Mehmet Gokmen},
  title = {A Character Segmentation Method to Increase Character Recognition Accuracy for Turkish License Plates},
  journal = {Mathematics and Computer Science},
  volume = {6},
  number = {6},
  pages = {92-104},
  doi = {10.11648/j.mcs.20210606.13},
  url = {https://doi.org/10.11648/j.mcs.20210606.13},
  eprint = {https://article.sciencepublishinggroup.com/pdf/10.11648.j.mcs.20210606.13},
  abstract = {Automatic License Plate Recognition is a computer vision technology that provides a way to recognize the vehicle's license plates without direct human intervention. Developing Automatic License Plate Recognition methodologies is a widely studied topic among the computer vision community to increase the accuracy rates. Automatic License Plate Recognition systems include image acquisition and character segmentation phases. Although there are many studies, the research in character segmentation and improving recognition accuracy remains limited. The lack of an international standard for license plates and the misinterpretation of ambiguous characters are challenging problems for Automatic License Plate Recognition systems. Several academic works have shown that the ambiguous character problem can be overcome by using a second model that contains only these characters. In this study, we propose a new methodology to reduce the character recognition errors of Automatic License Plate Recognition systems. One of the reasons for the low accuracy rates is the problem of ambiguous characters. In most studies using OCR, it was observed that a single model was used for alphanumeric characters during the recognition phase. Instead of using a single model, using separate models for letters and digits will improve the recognition process and increase accuracy. Therefore, we determined whether the characters are letters or numbers, and we expressed the license plates in the form of letters - digits. The method suggested for segmenting blobs worked with an accuracy of 96.12% on the test dataset. The method recommended for generating letter-digit expressions for the license plates worked with an accuracy of 99.28% on the test dataset. The proposed methodology can work only on Turkish license plates. In future studies, we will expand our method by using the license plate dataset of a different country.},
 year = {2021}
}

Copy | Download

TY  - JOUR
T1  - A Character Segmentation Method to Increase Character Recognition Accuracy for Turkish License Plates
AU  - Gulsum Cigdem Cavdaroglu
AU  - Mehmet Gokmen
Y1  - 2021/12/31
PY  - 2021
N1  - https://doi.org/10.11648/j.mcs.20210606.13
DO  - 10.11648/j.mcs.20210606.13
T2  - Mathematics and Computer Science
JF  - Mathematics and Computer Science
JO  - Mathematics and Computer Science
SP  - 92
EP  - 104
PB  - Science Publishing Group
SN  - 2575-6028
UR  - https://doi.org/10.11648/j.mcs.20210606.13
AB  - Automatic License Plate Recognition is a computer vision technology that provides a way to recognize the vehicle's license plates without direct human intervention. Developing Automatic License Plate Recognition methodologies is a widely studied topic among the computer vision community to increase the accuracy rates. Automatic License Plate Recognition systems include image acquisition and character segmentation phases. Although there are many studies, the research in character segmentation and improving recognition accuracy remains limited. The lack of an international standard for license plates and the misinterpretation of ambiguous characters are challenging problems for Automatic License Plate Recognition systems. Several academic works have shown that the ambiguous character problem can be overcome by using a second model that contains only these characters. In this study, we propose a new methodology to reduce the character recognition errors of Automatic License Plate Recognition systems. One of the reasons for the low accuracy rates is the problem of ambiguous characters. In most studies using OCR, it was observed that a single model was used for alphanumeric characters during the recognition phase. Instead of using a single model, using separate models for letters and digits will improve the recognition process and increase accuracy. Therefore, we determined whether the characters are letters or numbers, and we expressed the license plates in the form of letters - digits. The method suggested for segmenting blobs worked with an accuracy of 96.12% on the test dataset. The method recommended for generating letter-digit expressions for the license plates worked with an accuracy of 99.28% on the test dataset. The proposed methodology can work only on Turkish license plates. In future studies, we will expand our method by using the license plate dataset of a different country.
VL  - 6
IS  - 6
ER  - 

Copy | Download