Fuzzy Logic-Based Classification of Crescent Moon Images Using Brightness and Thickness Parameters Yudhiakto Pramudya, Kartika Firdausy, Adi Jufriansah, Okimustava, Itsnaini Irvina Khoirunnisa, Bayu Krisna Murti, Rihmah Alifah Hidayah, Murinto
Universitas Ahmad Dahlan, Universitas Muhammadiyah Maumere
Abstract
Accurate observation of the crescent moon holds significant importance in both astronomical research and calendrical determinations, such as the Islamic lunar calendar. However, detecting the thin crescent under bright sky conditions remains a challenging task due to its low contrast and subtle visual features. This study presents a novel fuzzy logic-based approach for classifying the visibility of crescent moon images obtained at the Observatorium Universitas Ahmad Dahlan (UAD). The methodology involves image preprocessing and fuzzification based on two key perceptual parameters: brightness and arc thickness. Brightness is categorized into low, medium, and high, while thickness is classified as thin, medium, and thick. A fuzzy inference system, developed in Python, is employed to evaluate and classify the images using triangular membership functions. The defuzzification process determines the visibility status of the crescent moon in one of three categories: not visible, maybe visible, and highly visible. Based on the analysis of 61 crescent moon images, 18 were classified as not visible, 11 as maybe visible, and 32 as highly visible. These results represent an important step toward the development of image recognition procedures using machine learning. Such recognition is essential for assessing crescent moon visibility under challenging daylight conditions and contributes to enhancing observational consistency in crescent moon detection for both calendrical and astronomical purposes.
