Determination of Real-Time Positions of a Seagoing Vessel Based on Real-Time Video

Abstract

This thesis investigates the possibility of determining the position of a seagoing vessel based solely on the video streams from multiple cameras. The problem is addressed in relation to offshore crane operations where there is relative movement between supply vessel and crane which is mounted on a rig or platform. The thesis starts by describing the offshore crane scenario and states the required properties of a prototype system. The video analysis starts by deriving an algorithm for finding the vessel in the video stream. Other algorithms for video tracking, finding corresponding points and estimating position and attitude of the vessel's cargo deck are presented. The system is implemented as a C++ application and is tested in a downscaled environment with a realistic, wave-simulated ship model. The measured positions are compared with the real positions and the accuracy of the signals are analyzed.