A comparison of calibration methods and system configurations of underwater stereo-video systems for applications in marine ecology

Abstract

Stereo-video systems are used by marine ecologists to count and measure the size of aquatic organisms. Most marine ecologists are not experts in underwater photogrammetry and are not necessarily aware of some of the key design and implementation issues that influence the accuracy and precision of the measurements. Two factors that can influence the measurement accuracy and precision of a stereo-video system are the calibration strategy and the physical orientation of the system. We compare the accuracy and precision of length measurements made by systems calibrated using two regularly used calibration approaches. We also compare length measurements recorded from cameras with base separations of 150 mm, 400 mm, and 800 mm. A three-dimensional (3D), network calibration method using a purpose built calibration cube was used to determine key calibration and camera orientation parameters (e.g., base separation, focal length, and lens distortions) with much improved precision in comparison to a two-dimensional method using either an A3 or A4 planar calibration pattern. As a result, measurements made with a 3D cube displayed improved accuracy and precision compared to either the A3 or A4 planar calibration pattern across a range of typical operational distances. The distance between the cameras on the base bar (base separation) affects the accuracy and precision of measurements with the 800 mm system producing length estimates of improved accuracy and precision than the systems with narrower baselines. The 150 mm system resulted in measurements with poor precision and accuracy, especially for measurements made at distances greater than six meters.