Implementation Analysis of GLCM and Naive Bayes Methods in Conducting Extractions on Dental Image

International Conference on Informatics, Engineering, Science and Technology

计算机科学;工业技术

Wijaya, E.^1

Informatics Engineering Program, Universitas Komputer Indonesia (UNIKOM), Jl. Dipatiukur No. 112-116, Coblong, Lebakgede, Bandung, Indonesia^1

关键词: Characteristic extraction;    Dental images;    Gray level co occurrence matrix(GLCM);    Gray level difference methods;    High-accuracy;    Implementation analysis;    Original images;    Texture feature extraction;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/407/1/012146/pdf
DOI  :  10.1088/1757-899X/407/1/012146

来源: IOP

【 摘 要 】

In a previous study identifying human tongue disease by analyzing texture feature extraction using the gray level difference method combined with naive bayes method obtained an accuracy of 81.1%. Another study using the same method, the naive bayes method combined with the Gray-Level Co-Occurrence Matrix (GLCM) method in different cases also received a high accuracy value of 96%. It proves that both methods can extraction texture of image well. This study aims to analyze what affects the accuracy of the accuracy of both GLCM and Naive Bayes methods in perforMing characteristic extraction of dental abnormalities. Dental abnormalities or anomalies used in this study were GeMination, Microdontia, and Makrodontia taken from the original image by simply showing the teeth from the front. From the test results, obtained an accuracy of 66.6%, which proves that the method of Naive Bayes is not good enough in doing the classification of dental abnormalities. This is because the data extracted characteristic texture of dental abnormalities with GLCM method has a matrix that is quite close distance between classes. So the image classification of tooth disorders using Naive Bayes method becomes less good when doing the classification.