Comparative Analysis of K-Means Clustering and Otsu Thresholding Methods in Digital Fingerprint Image Segmentation

Rifqi Arrafi (Universitas Islam Negeri Sumatera Utara)
Dedek (Universitas Islam Negeri Sumatera Utara)
Fauhan Alfarizi Saragih (Universitas Islam Negeri Sumatera Utara)
Muhammad Rayhans Adrians (Universitas Islam Negeri Sumatera Utara)
M. Haraji (STMIK TIME)

JITCoS : Journal of Information Technology and Computer System
Vol. 2 No. 1 (2026): June

Image segmentation is an important stage in digital image processing that aims to separate the main object from the background so that image information can be analyzed more accurately. This study compares the performance of Otsu Thresholding and K-Means Clustering methods in digital fingerprint image segmentation using a fingerprint image from the FVC2000 dataset. Prior to segmentation, preprocessing was performed using Gaussian Blur and Histogram Equalization to improve image quality. The evaluation was conducted using Mean Squared Error (MSE) and Peak Signal-to-Noise Ratio (PSNR). Experimental results on 20 fingerprint images from the FVC2000 dataset showed that the K-Means Clustering method achieved a lower average MSE value (5162.54) and a higher average PSNR value (11.34 dB) compared to Otsu Thresholding, which obtained an average MSE value of 8593.21 and an average PSNR value of 9.56 dB. Furthermore, Otsu Thresholding demonstrated significantly faster computational performance. These results indicate that K-Means provides better segmentation quality, whereas Otsu Thresholding is more computationally efficient.