Bullet Image Classification using Support Vector Machine (SVM)

2015 International Conference on Mathematics, its Applications, and Mathematics Education

数学;教育

Dwi Ratna, S.^1 ; Setyono, Budi^1 ; Herdha, Tyara^1

Department of Mathematics, InstitutTeknologiSepuluh November Campus ITS, Sukolilo, Surabaya

60111, Indonesia^1

关键词: Accuracy rate;    Defect detection;    Human errors;    Image defects;    Manual inspection;    Pre-processing;    Test data in;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/693/1/012009/pdf
DOI  :  10.1088/1742-6596/693/1/012009

学科分类：发展心理学和教育心理学

来源: IOP

【 摘 要 】

All Quality control of the bullet is usually done manually. Manual inspection process has some disadvantages for instance human error factor and the time inspection are relatively longer. In this study, defect detection and classification of bullets by using Support Vector Machine (SVM)has been done. The stages of classification of scheme bullets in this study include pre-processing, feature extraction with Principal Component Analysis (PCA) and classification of bullets by using SVM normalized. Performance of the proposed classification of scheme was tested using 80 images of bullets in which 40 images free of defects and 40 images with defect. The results show that the scheme can classify images of bullets with a 90% accuracy rate with test data in the form of 10 images free-defect and 10 image defects.