Evidence record · UDINUS · SDGs
- Unit
- Rektorat UDINUS
- Period
- 2024
- Goals
- Record
- f270949e-450d-4940-bb12-b6635b68936f
- Published
- 2026-10-30
Universitas Dian Nuswantoro (UDINUS), with Alzami, Farrikh; Winarno, Sri; Naufal, Muhammad; and Al Azies, Harun as activity participants, conducted research on integrating the Gaussian Mixture Model (GMM) with Contrast Limited Adaptive Histogram Equalization (CLAHE) to enhance image contrast for driver drowsiness detection based on facial images.
The activity addressed the limitation of traditional CLAHE, which uses a fixed clip limit that may not adapt effectively to varying image characteristics. UDINUS examined a CLAHE+GMM approach that dynamically determines the optimal contrast limit through GMM clustering. The stated aim was to achieve more adaptive and effective contrast enhancement for the facial images used in driver drowsiness detection.
The research evaluated the methods using Mean Squared Error (MSE), Peak Signal-to-Noise Ratio (PSNR), Signal-to-Noise Ratio (SNR), and computation time. These metrics provided the basis for assessing the contrast-enhancement methods. Experimental results indicated that using two clusters in CLAHE+GMM significantly improved MSE, PSNR, and SNR. Visual histogram comparisons and sample images further showed that CLAHE+GMM with two clusters closely aligned with the grayscale base image, supporting stable and consistent enhancement.
UDINUS therefore documented the CLAHE+GMM approach, particularly with two clusters, as a promising solution for image-contrast enhancement in driver drowsiness detection applications. The source does not provide additional institutional details about the research setting, implementation, subsequent application, or its relationship to a specific development goal.