Automated marker detection and time series analysis for through-water photogrammetry

Abstract. Through-water photogrammetry for surveys of Underwater Cultural Heritage (UCH) is a problem addressed by scholars from multiple perspectives. In addition to light attenuation as a function of distance travelled by light radiation through the liquid and variation of water optical characteristics across short lenghts, undulation of the free surface and magnitude of refraction (which varies at every point of each acquired image) also come into account. Our research aims to implement an automatic analysis method to estimate a priori — and correct a posteriori — the camera behaviour, focussing on surveys of objects in shallow waters. We proposed and tested a mathematical model to represent the effects of refraction across different water levels and states of motion. In addition to acquiring image pairs with two consumer-grade cameras, we recorded short video sequences using a high-speed, highsensitivity monocular camera to enable time-series analysis of the phenomenon.