An innovative and cost effective IOT system for character recognition of vehicle license plates

Abstract

Tracking the stolen vehicles, parking vehicle identification and parking the vehicles in the restricted areas is a big challenge nowadays. The main way of addressing these challenges is to recognize them through the number of plates embedded with technology. The task is therefore to recognize the position and the character on the number plate of the car. The current research article proposed an innovative cost-effective dashboard to identify the number of plates by capturing the image using a camera that is plugged with Raspberry Pi. The character recognition on the number plate is done by using K-Nearest Neighbors (KNN) algorithm. The images of the number plate are captured using raspberry pi cam and the numbers from the plate are extracted using Open CV. These identified number of plates will be trained using machine learning (ML) models. Based on the mean and variance the Euclidian distance will be calculated for training purposes. After training the large available datasets with all the unique number plates the trained data will be in the form of some encrypted format. So by inputting the new image (number plate) the features will be extracted and based on the calculation of the distance the rank will be allocated with the new plate and gets the nearest plate with character recognition the number will be extracted.