Comparison of the Effect of Filtering Methods on Fingerprint Accuracy Using the AFIS System on the BRAIL Attendance Machine.

Abstract

Fingerprint-based attendance devices are a secure and efficient solution, but their matching performance is highly dependent on image quality. Noise caused by finger conditions and sensor limitations can reduce the accuracy of the system, so it is necessary to improve the quality of fingerprint images. This study compares four image enhancement methods, namely Gaussian, Median, Bilateral, and CLAHE filters, using two data categories to determine the most suitable method for the BRAIL laboratory attendance machine. The evaluation was based on the increase in matching confidence. The test results show that the Gaussian filter increased confidence by 23.81% in the first data category, but only 1.02% in the second data category. The Median filter produced a confidence increase of 21.37% in the first data category and the highest increase of 7.59% in the second data category. Based on these results, the Median filter produced the highest cumulative confidence due to its ability to maintain fingerprint ridge details and its resistance to noise. Therefore, the Median filter is recommended as the most effective method for improving fingerprint image quality in fingerprint-based attendance systems.