Young lunar crescent detection based on video data with computer vision techniques

Abstract

More than one and a half billion people worldwide use a calendar based on the young lunar crescent apparition. In the Hijri calendar, it is crucial to recognize the appearance of the lunar crescent for the first time after conjunction, especially for the three critical months of Ramadan, Shawwal, and Dzulhijjah. This work uses video data to be processed using computer vision algorithms to identify the young lunar crescent’s appearance. The lunar crescent will be captured using Gaussian Blur and Adaptive Thresholding. Image processing techniques, such as frame extraction from video, pre-processing images, and detection algorithms utilizing Circular Hough Transform (CHT), are all implemented using the OpenCV package. We chose ten observations as a sample provided by the Meteorological, Climatological, and Geophysical Agency (MCGA). The computation time of the proposed model is relatively faster than the frequency of the frames displayed per second on the video. The program’s accuracy in distinguishing among frames with and without the lunar crescent is quite good. We conclude that the proposed model can accurately and quickly detect when the young lunar crescent will appear.