Abstract
Advances in 3D scanning and reconstruction techniques, such as structure-from-motion, have resulted in an abundance of increasingly-detailed 3D habitat models. However, many existing methods for calculating structural complexity of these models use 2.5D techniques that fail to capture the details of truly 3D models with overlapping features. This paper presents novel algorithms that extend traditional rugosity metrics to generate multi-scale rugosity maps for complex 3D models. Models are repeatedly subdivided for local analysis using multiple 3D grids, which are jittered to smooth results and suppress extreme values from edge cases and poorly-fit reference planes. A rugosity-minimizing technique is introduced to find optimal reference planes for the arbitrary sections of the model within each grid cell. These algorithms are implemented in an open-source analysis software package, HabiCAT 3D (Habitat Complexity Analysis Tool), that calculates and visualizes high-quality 3D rugosity maps for large and complex models. It also extends fractal dimension and vector dispersion complexity metrics, and is used in this paper to evaluate and discuss the appropriateness of each metric to various coral reef datasets.