Melinda, Maria (2015) Identifikasi cacat amunisi dengan menggunakan pengolahan citra digital dan linear discriminant analysis (LDA). Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
1211100029-Undergraduated_Thesis.pdf - Published Version Download (3MB) | Preview |
Abstract
Pengecekan pada produksi amunisi secara manual
memiliki kemungkinan besar terjadi kesalahan karena
dipengaruhi subjektivitas dan tingkat kelelahan visual. Dalam
tugas akhir ini penulis melakukan deteksi kecacatan pada amunisi
dengan menggunakan pengolahan citra digital dan metode Linear
Discriminant Analysis (LDA). LDA merupakan salah satu metode
untuk melakukan reduksi dimensi dan ekstraksi fitur. Adapun
tahapan dari proses deteksi kecacatan pada tugas akhir ini
meliputi proses akuisisi citra, proses pra-pengolahan, proses
ekstraksi fitur dengan LDA dan klasifikasi citra dengan
menggunakan jarak Euclidean ternormalisasi. Proses akuisisi citra
dilakukan pada 80 citra amunisi hasil pengambilan gambar, yang
terdiri dari 40 citra baik dan 40 citra cacat. Beberapa proses prapengolahan yang dilakukan adalah proses segmentasi dengan
thresholding Otsu, proses cropping, scalling, grayscalling dan
ubah dimensi. Berdasarkan hasil pengujian sistem deteksi
kecacatan dengan metode Linear Discriminant Analysis (LDA)
pada tugas akhir ini dapat mengenali citra amunisi dengan tingkat
akurasi sebesar 72% pada 10 data pelatihan baik dan cacat
dengan panjang fitur 400 dengan waktu komputasi terbaik untuk
pelatihan adalah 3 menit 8,7 detik.
==============================================================================================================
Manually checking ammunition has most likely occurred
due to the subjectivity of the error and visual fatigue. In this final
project the author to detect defects in the ammunition by using
digital image processing and methods of Linear Discriminant
Analysis (LDA). LDA is one method that perform dimension
reduction and feature extraction. The stages of the process of
detection of defects in this final project includes image
acquisition process, the process of pre-processing, feature
extraction process with LDA and image classification using
normalized Euclidian distance. Image acquisition process
performed on 80 shots ammunition image, which consists of 40
good image and 40 defect image. Some of the pre-processing is
performed by thresholding Otsu segmentation process, the
process of cropping, scaling, grayscalling and change
dimensions. Based on test results defect detection system with
Linear Discriminant Analysis (LDA) method in this paper can
recognize the image of ammunition with an accuracy rate of 72%
to 10 good and defect training data with a length of 400 features
with the best computing time for training is 3 min 8,7 sec.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Additional Information: | RSMa 006.42 Mel i |
| Uncontrolled Keywords: | Amunisi, Thresholding Otsu, LDA, Jarak Euclidean |
| Subjects: | Q Science > QA Mathematics |
| Divisions: | Faculty of Mathematics and Science > Mathematics > 44201-(S1) Undergraduate Thesis |
| Depositing User: | Mr. Fandika aqsa |
| Date Deposited: | 09 Jun 2017 03:09 |
| Last Modified: | 09 Jun 2017 03:09 |
| URI: | http://repository.its.ac.id/id/eprint/41550 |
Actions (login required)
 |
View Item |