Enhancement for License Plate Recognition using Image Super Resolution technique
Abdelsalam Hamdi Abdelaziz, Yee Kit Chan, Voon Chet Koo · 2021 International Conference on Electrical, Communication, and Computer Engineering (ICECCE) · 2021
License Plate Recognition (LPR) is one of the important applications of AI, however due to image blurring and low resolution, the Optical Character Recognition (OCR) models are unable to recognise the text accurately. Therefore, we propose a deblurring and image super resolution model for image recovery using deep learning and convolutional neural networks. Our results show an improvement of the Peak Signal to Noise Ration (PSNR) from 30.182 to 31.696. It is shown that adding a deblurring and image super resolution model can improve the OCR accuracy up to 62% from 12% and the average error rate from 6.1 to 2.6 per image.