EAPSI: A Novel Approach to Subsea Length Estimation Using Cameras in Parallel and Open-Source Software.

Johnson Garrett B, Kaneohe HI

Investigators

Abstract

Since the dawn of ecological studies of coral reefs in the 1950s, scientists have documented lengths of fish populations which indicate the overall health of fishery stocks and their habitat. The traditional method for fish length estimations is diver-based, where a researcher observes fish underwater and records length estimations by hand. Due to the optical properties of water these estimations are difficult and subject to researcher biases. Generally, fish lengths are averaged and grouped into a few size classes (Ex. 5 - 10cm, 11 - 20cm, 21 cm+). To increase the accuracy underwater length estimates, stereovideo cameras can be used, where two cameras are run in parallel and software calculates lengths of objects of interest to the nearest millimeter, a magnitude of greater accuracy than diver-based visual census techniques. The problem with using these systems over traditional methods is the high cost, which prevents widespread use. Accurate fish length records are critical for fishery resource managers in determining the overall health of coral reef systems to ensure that they remain healthy for future generations. In this study I will use OpenCV open-source software tools and low-cost GoPro underwater video cameras to develop an underwater diver based stereovideo system at a fraction of the cost of currently available systems. In Taiwan, I will be working with Dr. Colin Wen of Tunghai University, an expert on current fish length estimation techniques. Additionally, my low cost system will be used in collaboration with Dr. Wen and Dr. Allen Chen of Academia Sinica, in support of a current survey of commercially important fishery species of Taiwan. This award, under the East Asia and Pacific Summer Institutes program, supports summer research by a U.S. graduate student and is jointly funded by NSF and the Ministry of Science and Technology of Taiwan.