Author(s): Jakke Kulovesi; Jonni Lehtiranta
Linked Author(s):
Keywords: No Keywords
Abstract: Ship crews typically measure level ice thickness visually. Such measurement is possible when a block of ice turns sideways and reveals its cross-section while aligned with a reference measure. These estimates are subjective and rely on the experience and skills of the observer. In addition, manually performed visual thickness estimates have a very limited resolution. Furthermore, the task of observing ice thickness is mentally taxing, especially when performed for extended periods of time. This paper presents a semi-automatic machine vision based solution to the aforementioned problems (subjectivity, resolution, and cognitive strain). This solution continuously records images of rotated ice blocks using two high resolution industrial cameras. This stereo configuration allows reliable 3D measurement of ice dimensions. Human input is needed for choosing feasible cases. The presented computer program then detects edges and automatically calculates ice thickness. The process is monitored and results verified by the human operator. Data was collected during sea trials of the ship S. A. Agulhas II in the Bay of Bothnia and semi-automatic measurement was conducted for data collected during the trials. Experiences during the sea trials and the actual measurement results are discussed. The results of this work form a ground truth which can be used for validating the results of future work of higher level of automation.
Year: 2014