Automated Vehicle Number Plate Recognition System using Stability Score and K-Means Clustering Algorithm

Due to the increase in vehicle usage, itis a challenging task to monitor, analyze the vehicles by a human for security purposes. There is a need for an automatic vehicle recognition system since various places nowadays have checkpoints for vehicles, to track the stolen vehicles, and to monitor traffic violations. The problem exists when the vehicle number plate is encountered in different formats, different scales, and illumination to number-plates. In the case of an indeterminate situation, identifying vehicle number plates in poor lighting conditions and worse traffic situations can be analyzed using an automatic vehicle number plate recognition system. The vehicle name board edge finding techniques are used to easily identify the vehicle number in the name board. A dataset with 200 license plates has been collected as training datasets for recognition, estimation, and identification, thus improving system accuracy of recognition when compared to existing works. The training input samples include images of vehicle number plates taken from the traffic department. The automated vehicle number recognition system is improvised in terms of accuracy by estimating stability score and using the k-means clustering algorithm.