Persian Car License Plate Recognition Using Deep Convolutional Neural Networks | ||
| AUT Journal of Electrical Engineering | ||
| دوره 57، Issue 2 (Special Issue)، 2025، صفحه 295-316 اصل مقاله (1.52 M) | ||
| نوع مقاله: Research Article | ||
| شناسه دیجیتال (DOI): 10.22060/eej.2024.23296.5601 | ||
| نویسنده | ||
| Hossein Noori* | ||
| Department of Electrical Engineering, Vali-e-Asr University of Rafsanjan, Rafsanjan, Iran. | ||
| چکیده | ||
| Car license plate recognition plays a very important role in intelligent transportation systems. In this paper, a new car license plate recognition algorithm is proposed. To recognize the car license plate, first, a new dataset in different light conditions is collected and labeled. Then a new convolutional neural network is proposed to find the position of a car license plate in an image, named plate position detection network (PPDN). The characters on the plate are separated by a proposed algorithm after the position of the plate is determined. Since the Iranian car license plate includes both digits and letters, two other convolutional neural networks are proposed: one for digit recognition, named digit recognition network (DRN), and the other for letter recognition, named letter recognition network (LRN). All three networks are trained with related datasets. Simulation results show that the PPDN reaches 99.86% training accuracy, 99.75% validation accuracy, and 99.53% test accuracy. Also, DRN reaches 100% training accuracy, 99.6% validation accuracy, and 99.51 % test accuracy. In addition, LRN reaches 100% test accuracy, 99.45% validation accuracy, and 99.05% test accuracy. Finally, the accuracy of the network for plate recognition on the test data is 99.21%. | ||
| کلیدواژهها | ||
| Deep Learning؛ Object Detection؛ Pattern Recognition؛ Intelligent Transportation System | ||
| مراجع | ||
|
| ||
|
آمار تعداد مشاهده مقاله: 1,060 تعداد دریافت فایل اصل مقاله: 1,193 |
||
| تعداد نشریات | 9 |
| تعداد شمارهها | 455 |
| تعداد مقالات | 5,773 |
| تعداد مشاهده مقاله | 8,414,155 |
| تعداد دریافت فایل اصل مقاله | 6,974,730 |