License Plate Image Recognition System Based on Firefly Algorithm and BP Neural Network

Abstract

Aiming at the defects of low license plate recognition accuracy of the traditional BP (Back Propagation) neural network, a license plate recognition algorithm based on FA (Firefly algorithm) and BP neural network is proposed. Firstly, image enhancement is performed on a license plate image captured by cameras, and then, the license plate image is grayed by the maximum method, average method, and weighted average graying to obtain the best image. Secondly, Hough transform is adopted for license plate correction, segmentation, and location. Finally, FA is introduced into BP neural network to recognize license plate characters. The results show that compared with the traditional BP neural network license plate recognition algorithm, the algorithm proposed in this paper improves the recognition accuracy of Chinese characters, letters, and the overall license plate recognition rate by 4.6%, 2%, and 6.4% respectively.