License plate recognition (LPR) is composed of license plate detection (LPD), character segmentation, and character recognition. In this paper, we put forward a license plate recognition (LPR) system. For license plate detection (LPD), a hybrid algorithm based on colour and shape is proposed (HLPD). For character segmentation, taking into account the horizontal preliminary information of the black and white histogram, a cross-zero search character segmentation algorithm based on the black and white histogram is suggested (CCZA). For character recognition, 10 sets of CNN models with different frameworks are proposed. The CCCP-BN-DP performs best when compared to the performance of the test set. The HLPD-CCZA-CCCP-BN-DP system is then applied to the public data set CCPD and experiments demonstrate the effectiveness of the LPR system.
License plate recognition system based on the difference of convolutional neural network framework
08.10.2022
722228 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
License Plate Recognition System Based on Neural Network
Springer Verlag | 2023
|Europäisches Patentamt | 2022
|Europäisches Patentamt | 2015
|LICENSE PLATE RECOGNITION SYSTEM AND LICENSE PLATE RECOGNITION METHOD
Europäisches Patentamt | 2020
|License plate recognition system and license plate recognition method
Europäisches Patentamt | 2020
|