PERBANDINGAN PERFORMA PENGGABUNGAN METODE LOCAL BINARY PATTERN HISTOGRAM (LBPH) DENGAN DETEKSI TEPI BERBASIS GRADIEN PADA SISTEM PENGENALAN WAJAH MANUSIA

RAKHMAN, MUHAMMAD FAJAR and Fachrurrozi, Muhammad and Rachmatullah, Muhammad Naufal (2022) PERBANDINGAN PERFORMA PENGGABUNGAN METODE LOCAL BINARY PATTERN HISTOGRAM (LBPH) DENGAN DETEKSI TEPI BERBASIS GRADIEN PADA SISTEM PENGENALAN WAJAH MANUSIA. Undergraduate thesis, Sriwijaya University.

	Text RAMA_55201_09021381722089.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (2MB) | Request a copy
	Text RAMA_55201_09021381722089_TURNITIN.pdf Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (8MB) | Request a copy
	Text RAMA_55201_09021381722089_2107109003_0001129204_01_front_ref.pdf - Accepted Version Available under License Creative Commons Public Domain Dedication. Download (896kB)
	Text RAMA_55201_09021381722089_2107109003_0001129204_02.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (153kB) | Request a copy
	Text RAMA_55201_09021381722089_2107109003_0001129204_03.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (200kB) | Request a copy
	Text RAMA_55201_09021381722089_2107109003_0001129204_04.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (692kB) | Request a copy
	Text RAMA_55201_09021381722089_2107109003_0001129204_05.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (437kB) | Request a copy
	Text RAMA_55201_09021381722089_2107109003_0001129204_06.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (65kB) | Request a copy
	Text RAMA_55201_09021381722089_2107109003_0001129204_07_ref.pdf - Bibliography Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (71kB) | Request a copy
	Text RAMA_55201_09021381722089_2107109003_0001129204_08_lam.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (46kB) | Request a copy

Abstract

The use of the LBPH method in face recognition is the most efficient method compared to previous methods such as Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and Elastic Bunch Graph Matching (EBGM). However, the performance of the LBPH method can be further improved with a combination of gradient-based edge detection methods. Sobel, Prewitt, and Robert's edge detection method serves as a provider of gradient values for each image pixel that will help in the LBP operation process. In this study, three classification processes were carried out using each combination of methods, namely Sobel-LBP, Prewitt-LBP, and Robert-LBP on the “Yale Faces Database” dataset. Each accuracy value generated is 89%, 87%, and 84%, while the LBPH method without using edge detection is 76%. The test results show that the combination of gradient-based edge detection methods with LBP operations can improve recognition performance with the highest accuracy value in Sobel-LBP, because the kernel size in the Sobel method is larger than that of Prewitt and Robert. However, the three edge detection methods have the same limitations, which are very sensitive to noise in the image.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Classification, Biometric, Face Recognition
Subjects:	Q Science > Q Science (General) > Q300-390 Cybernetics > Q325.5 Machine learning
Divisions:	09-Faculty of Computer Science > 55201-Informatics (S1)
Depositing User:	Muhammad Fajar Rakhman
Date Deposited:	22 Nov 2023 08:11
Last Modified:	22 Nov 2023 08:11
URI:	http://repository.unsri.ac.id/id/eprint/130895

Actions (login required)

View Item