Use Case of HPVT in Counting Stars: A Vision-Based Approach of Monitoring Natural Events

Counting stars could reveal the level of a cloudy sky at night, which provides important climate information. The monitoring of natural events, such as counting stars, can pose as a highly demanding task which requires an immense amount of time and dedication to be completed by a human observer. By using rich visual data from a video transmission system, namely HpVT, as the input, a specialized monitoring program utilizing computer vision techniques was developed to count stars under various night sky conditions. We present the design of a star counting framework and describe the counting process employed. We perform testing on the star counting framework using our own data set. The data set consists of 154 video footages (of 24-h period each) contain footages of the night sky which were recorded from various locations in different countries. This work shows a proof of concept that the number of stars observed can correlate well with climate information. We also discuss some of the issues faced and make recommendations for future projects.