IMAGE PROCESSING AND MACHINE LEARNING-BASED TOLL COLLECTION MODEL FOR INDIAN HIGHWAYS

Abstract

The road transport system plays a crucial role in fostering economic and societal growth. However, it faces numerous obstacles, such as traffic congestion, inefficient toll collection methods, and insufficient enforcement of traffic regulations. This study proposes a Distance-Based Toll Collection Model using Automatic Number Plate Recognition (ANPR) and machine learning techniques to address these problems. The model aims to enhance the efficacy, equity, and safety of toll collection while promoting optimal road network utilisation. The proposed system employs ANPR technology to automatically identify vehicles through licence plate recognition, thereby eliminating the need for manual toll collection and decreasing congestion at toll booths. The collection of real-time data enables efficient traffic monitoring and management. The distance-based toll collection model ensures a fair and equitable system by charging tolls based on the actual distance travelled by vehicles. This promotes cost effectiveness and encourages optimal road network utilisation. The application of machine learning techniques, particularly the Indian Number Plate Character Recognition (INPCR) model, improves the accuracy and dependability of number plate recognition, even for Indian number plates with varying designs. In terms of accuracy, precision, recall, F1 score, and Receiver Operating characteristic - Area Under Curve, the evaluation results are promising. The implementation of the proposed system contributes economically to fuel and time savings. By minimising toll booth wait times and optimising traffic flow, congestion and delays are reduced, resulting in substantial fuel savings and decreased emissions. While the proposed system offers several advantages, there are obstacles and research gaps that need to be addressed. Variability in licence plate designs, deviations from ideal image conditions, the inability to generalise to non-standard plates, and the need for real-time performance optimisation are examples of these challenges. Distance-Based Toll Collection Model utilising ANPR and machine learning techniques provides a comprehensive and effective solution to road transportation system challenges. The proposed system contributes to a sustainable and user-friendly road network by reducing congestion, enhancing safety, promoting cost-effectiveness, and enhancing services. Future research and development should concentrate on overcoming obstacles to improve the system's efficacy and dependability.