Analysis of Image Processing in Barcode Using the K-Nearest Neighbor (K-NN) Classification

Abstract

© 2020, World Academy of Research in Science and Engineering. All rights reserved. Barcode is a product introduction method. Each product has a different shape on the barcode. The system created contains a barcode recognition based on the image you have. This system aims to recognize products based on barcode characteristics. The system being studied can be used by users when shopping. The method used in this system for feature extraction is Hough Transform and the classification method used is k-NN. This research is expected to be able to classify barcode correctly and have good accuracy. This application uses distance parameters = 5cm, 10cm, and 15cm and uses angles = 0˚, 45˚, 90˚, 135˚ and 180˚. The best results are obtained from a distance of 10 cm with angles = 0, and compared with variations in the value of k. A value of k = 1, 3, 5 gets a 71% result which is the best accuracy result in this system. Based on the best results, therefore to improve the accuracy of the system, the best distance and angle are taken from the previous parameters. This system using the Hough Transformation feature extraction method and the k-NN class classification get an accuracy value of 100%, with a distance of taking 10 cm and an angle of 0°.