PENINGKATAN KUALITAS CITRA DENGAN METODE RETINEX UNTUK DETEKSI WAJAH PADA CITRA GELAP

SAVERO, MUHAMMAD AWADH and Fachrurrozi, Muhammad and Rachmatullah, Muhammad Naufal (2025) PENINGKATAN KUALITAS CITRA DENGAN METODE RETINEX UNTUK DETEKSI WAJAH PADA CITRA GELAP. Undergraduate thesis, Sriwijaya University.

	Image RAMA_55201_09021381924126_cover.jpeg - Cover Image Available under License Creative Commons Public Domain Dedication. Download (466kB)
	Text RAMA_55201_09021381924126.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (2MB) | Request a copy
	Text RAMA_55201_09021381924126_turnitin.pdf - Accepted Version Available under License Creative Commons Public Domain Dedication. Download (6MB)
	Text RAMA_55201_09021381924126_0222058001_0001129204_01_front_ref.pdf - Accepted Version Available under License Creative Commons Public Domain Dedication. Download (779kB)
	Text RAMA_55201_09021381924126_0222058001_0001129204_02.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (380kB) | Request a copy
	Text RAMA_55201_09021381924126_0222058001_0001129204_03.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (307kB) | Request a copy
	Text RAMA_55201_09021381924126_0222058001_0001129204_04.pdf - Accepted Version Restricted to Registered users only Available under License Creative Commons Public Domain Dedication. Download (514kB) | Request a copy
	Text RAMA_55201_09021381924126_0222058001_0001129204_05.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (630kB) | Request a copy
	Text RAMA_55201_09021381924126_0222058001_0001129204_06.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (17kB) | Request a copy
	Text RAMA_55201_09021381924126_0222058001_0001129204_07_ref.pdf - Bibliography Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (143kB) | Request a copy
	Text RAMA_55201_09021381924126_0222058001_0001129204_08_lamp.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (293kB) | Request a copy

Abstract

Facial images captured under low-light conditions often suffer from visual degradation, including low contrast, inadequate brightness, and loss of important details, which negatively impact face detection accuracy. This study aims to enhance the quality of low-light images using the Retinex method as a preprocessing step and to evaluate its effect on the performance of the YOLOv5 face detection algorithm. A total of 6,000 low-light facial images were used in the experiment. Image enhancement was assessed using brightness, contrast, entropy, Peak Signal-to-Noise Ratio (PSNR), Mean Squared Error (MSE), and Structural Similarity Index Measure (SSIM). The Retinex method successfully improved the average brightness to 104.57, contrast to 21.50, entropy to 391,715, and reduced the MSE to 8,985.14. However, the average PSNR and SSIM values—8.74 dB and 0.1359 respectively—indicate that some distortion remained in several images. The enhanced images were then used to train and test the YOLOv5 model, achieving the best performance at a batch size of 32 and 100 epochs, with a confidence score of 0.93.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Peningkatan Kualitas, Retinex, YOLOv5, Citra Wajah Gelap.
Subjects:	Q Science > Q Science (General) > Q334-342 Computer science. Artificial intelligence. Algorithms. Robotics. Automation.
Divisions:	09-Faculty of Computer Science > 55201-Informatics (S1)
Depositing User:	Muhammad Awadh Savero
Date Deposited:	03 Sep 2025 08:15
Last Modified:	03 Sep 2025 08:15
URI:	http://repository.unsri.ac.id/id/eprint/183565

Actions (login required)

View Item