Camera array modeling for large scale coral reef surveillance through area calibration of underwater camera

Abstract

Video monitoring is one of the fastest and most efficient method of coral surveillance, however, spatial information of the assessed coral reefs is not readily available from video data alone. The physical area of images are calculated using the relationship of meter-to-pixel scale factor and depth. Using this information, a three-camera array was designed to improve the benthic coverage of surveillance per one sweep such that it produce 40% overlap between their captured images. This image overlap ensures that image mosaicking technique by Corpuz [6] would work on the series of captured images. When implemented at an altitude of 1.2 meters from the reef floor, the error from the theoretical calculations was 7.5% between
cameras 1 and 2 and 12% between cameras 2 and 3. The camera array increased the scope of the coral data in one sweep by 70% compared to when a single camera was used.