Automatic IOT and Machine Learning–Based Toll Collection System for Moving Vehicles

The proposed work aims to develop an automated toll collection system for charging vehicles passing through a toll plaza. In India, the existing manual toll collection system is slow and leads to heavy traffic congestion and long queues at toll plazas. This causes significant waiting time, fuel wastage, increased pollution, and reduced highway speeds, negatively affecting the quality of food products, especially perishable items like milk. To address these challenges, an effective solution is presented here, focusing on localizing the license plates of vehicles. This automatic license plate recognition approach utilizes various image processing techniques to read the number plate of a moving car. Initially, the captured image is converted to grayscale to optimize memory consumption and processing speed, by selecting a single plane from the RGB (red, green, blue) channels. A camera captures the image of the moving vehicle from a fixed distance throughout the process. To enhance the image quality, morphological filters are applied, followed by optical character recognition (OCR) using a convolutional neural network (CNN). A database has been created and linked to test the performance of the prototype toll collection system. This database contains information such as the number of vehicles, vehicle owners’ names, unique identification numbers, mobile numbers, and linked bank account balances. The recognized characters from the license plate are compared with the database for matching. Once a match is found, the vehicle proceeds through the toll collection process. This involves an Internet of Things (IoT) integration, where barriers open and close automatically based on the system’s response. The main advantages of this system include fast response times and efficient detection. The results demonstrate a significant reduction in vehicle waiting time, queue length, fuel wastage, and pollution at toll plazas.