Open Access2D Detection Model of Defect on the Surface of Ceramic Tile by an Artificial Neural Network
Author(s) -
B Mariyadi,
Norma Latif Fitriyani,
Taufik Roni Sahroni
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1764/1/012176
Subject(s) - artificial intelligence , artificial neural network , backpropagation , computer science , computer vision , rgb color model , feature extraction , pattern recognition (psychology) , feature (linguistics) , preprocessor , process (computing) , philosophy , linguistics , operating system
The ceramic tile visual inspection process is divided into three parts: texture classification, color classification, and surface defect detection. In its application in the industry is a difficult process, because it is done manually involving many workers and done in a noisy environment with differences in temperature and humidity. This study emphasized the quality control based on the visual inspection automation system on the detection of a defective type of ceramic surface. The process performed is capturing an image, image processing, feature extraction, training, testing, and classification. In the preprocessing process are image resizing, RGB color conversion, segmentation and feature extraction. Features extraction using Gray Level Co-occurrence Matric (GLCM). The generated feature will be an artificial neural network (ANN) input in training and testing using Matlab 2013a to detect more than one type of surface defect with good accuracy. The artificial neural network training uses backpropagation with network architectures 14 input features, 27 hidden layers, 1 output. The learning rate used 0.001, 75 data training, and 23 data testing. The position and type of defect can be detected with 83% accuracy and error rate of 17%. Maximum time detection 1,06 seconds and minimum time detection 0,1 seconds. therefore by using automation system inspection, inspection error caused humans could be ignored and increased quality and productivity in production.