DESIGN OF AUTOMATIC NUMBER PLATE RECOGNITION (ANPR) SYSTEMS BASED ON K-NN MACHINE LEARNING AN ON THE RASPBERRY PI EMBEDDED SYSTEM

• Main Author: Sugeng
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/41721
• Institution: Institut Teknologi Bandung

Abstract—Research about number plate recognition or Automatic Number Plate Recognition (ANPR) mostly to be done by researchers to produce an recognition that has high accuracy. Some methods of the recognition are carried out likes as recognition to edge detection and morphology, relationship between objects analysis, machine learning and deep learning. In research was developed the ANPR system based on machine learning K-NN as a character recognition method. The method of connected component analysis used to localize number plates. The system was developed also add an artificial intelligence to perform optimization of character recognition that is less precise based on the position of the character group in the number plate. The ANPR system develop was designed to be able to carry out the introduction of two-wheeled and four-wheeled vehicle license plates implemented on an Raspberry pi embedded system. The ANPR system was also developed to use in the parking management system. In this research the recognition number plates are limited to private number plates in Indonesia. After testing, the system made capable of recognizing the vehicle number plate properly on vehicles that have standard license plates according of the National Police regulations, both in the font type and in the number plate writing format, both in the font type and the number plate writing format. From 103 vehicle images that were sampled, 3 vehicles cannot be recognized perfectly.