Sistem Klasifikasi Mutu Keramik Menggunakan Fitur Glcm (Grey Levelco-Occurrence Matrix) Dengan Membandingkan Metode Klasifikasi Svm Dan Bpnn

Pangestu, Dwiky Bintang (2021) Sistem Klasifikasi Mutu Keramik Menggunakan Fitur Glcm (Grey Levelco-Occurrence Matrix) Dengan Membandingkan Metode Klasifikasi Svm Dan Bpnn. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

Text 10311710000060_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2023. Download (3MB) | Request a copy

Abstract

Keramik di Indonesia menurut Kementerian Perindustrian
menilai bahwa akan mampu bersaing dipasar perdagangan Internasional, sebab proses produksi maupun pemasaran sudah melakukan penerapan teknologi yang canggih ke beberapa belahandunia. Industri ini diprediksi
akan terus maju dan mengkilap sebab, Indonesia memiliki keunggulan dan ketersediaan potensi sumber daya bahan.
Permasalahan yang sering terjadi yaitu tekstur surface keramik yang mirip grade 1,grade 2 dan grade 3, banyak pabrik masih menggunakan penglihatan pada operatornya. Oleh karena untuk mencegah human eror itu dilakukan penelitian deteksi kerusakan keramik dengan menggunakan metode GLCM (Gray Level Co- Occurrence Matricies) untuk mengkonversi data citra sehingga menghasilkan data
numerik. Lalu dilanjutkan dengan mencari perhitungan akurasi menggunakan algoritma Neural Network Klasifikasi dan SVM, keramik
nantinya akan membantu dalam mendeteksi kualitas keramik
berdasarkan grade nya. Kernel SVM yang digunakan pada penelitian ini yaitu kernel linear, RBF, polynomial dan menggunakan metode BPNN.Pengujian dilakukan dengan beberapa uji coba berdasarkan nilai d dan θ pada
ekstraksi fitur GLCM dan berdasarkan kernel SVM dan BPNN yang digunakan. Setelah dilakukan pengujian didapatkan nilai d dan θ terbaik yaitu nilai d=1 dan θ=0⁰-135⁰ pada metode BPNN. Kemudian nilai terbaik dari d dan θ tersebut digunakan untuk mendeteksi mutu keramik secara realtime berdasarkan sisi pengambilan gambar objek dan didapatkan nilai akurasi 95% dengan akurasi sistem dalam mengklasifikasi mutu keramik dengan rata-rata waktu komputasi sebesar 1,2 detik.
=======================================================================================================
Ceramics in Indonesia according to the Ministry of Industry assesses that it will be able to compete in the international trade market, because the production and marketing process has been applying advanced technology to several parts of the world. This industry is
predicted to continue to advance and shiny because, Indonesia has the advantage and availability of potential material resources. The problem that often occurs is the texture of ceramic surfaces that are similar to grade 1, grade 2 and grade 3, many factories still use
vision on the operator. Therefore, to prevent human error was conducted ceramic damage detection research using GLCM (Gray Level Co-Occurrence Matricies) method to convert image data so as to generate numerical data. Then followed by finding accuracy calculations using Neural Network Classification and SVM algorithms,ceramics will later help in detecting the quality of ceramics based on their grade. Svm kernels used in this study are linear kernel, RBF, polynomial and bpnn method. Testing is conducted with multiple trials based on d and θ values on GLCM feature extraction and based on svm kernel and BPNN used. After testing, the best d and θ values are d=1 and
θ=0-135 in bpnn method. Then the best value of d and θ is used to detect the quality of ceramics in realtime based on the shooting side of object and obtained an accuracy value of 95% with the accuracy ofthe system in classifying the quality of ceramics with an average computing time of 1.2 seconds.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Keramik, Klasifikasi , GLCM, BPNN, SVM, Ceramics, Classification , GLCM, Neural Network, SVM, BPNN
Subjects:	Q Science > QA Mathematics > QA336 Artificial Intelligence
Divisions:	Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User:	Dwiky Bintang P
Date Deposited:	26 Aug 2021 14:48
Last Modified:	26 Aug 2021 14:48
URI:	http://repository.its.ac.id/id/eprint/89792

Actions (login required)

View Item