A Study of a Sobel Edge Detection and SVM Based ALPR System on an ARM Single-Board Microcomputer

License plate is the unique identification of the vehicle, and Automatic license plate recognition (ALPR) systems have emerged as one of the distinctive applications of Image Processing, representing a typical paradigm of optical character recognition (OCR). However, traditional license plate number recognition systems often encounter challenges such as high costs, large sizes, inefficient CPU resource utilization on computer platforms. Utilizing a single-board computer can effectively address these issues, offering more efficient resource utilization and satisfactory task completion. The Linux-based Acorn RISC Machine (ARM) embedded single-board system emerges as a suitable platform for this purpose. This paper primarily investigates an ALPR system based on the open-source image processing library Open Computer Vision (OpenCV), adapted for deployment on an ARM single-board microcomputer, referred to as the ARM-ALPR system in this study. The research encompasses license plate image preprocessing, character segmentation, character recognition, and cross-platform system migration, etc. The proposed methodology employs the Sobel edge detection algorithm and morphological algorithms to delineate license plate outlines, projection techniques for character segmentation, and employs a “one-to-many” recognition approach within the Support Vector Machine (SVM) classifier to enhance accuracy. Additionally, to improve system operability and aesthetics, PyQt5 is employed to design the GUI interface. Finally, the programming code is ported to the Raspberry Pi to realize the license plate recognition system. The Raspberry Pi-based license plate recognition system exhibits cross-platform compatibility and easy portability, capable of running on different hardware platforms with high flexibility and scalability. Therefore, the outcomes of the implementation yield accurate results, indicating its promising potential for a broader range of applications.