Sensor-Vision Fusion using Kalman filter
The cultural heritage of European nations is spread all across the continent. While things on
land are easily accessible, maritime heritage, such as shipwrecks or ancient underwater cities,
is often unreachable for tourists. Project i-MareCulture aims at bringing the underwater cultural
heritage within digital reach of the wide public by implementing virtual visits, serious
games with immersive technologies and underwater augmented reality. The thesis focuses on
augmented reality within this project. It is dealing with various problems, such as bad visibility,
lack of flat surfaces, and various forces interacting with the device. Therefore, the device
cannot rely only on the camera itself, but it has to combine the image with sensors, mainly accelerometer
and gyroscope. The goal of the thesis is to add the data from the sensors together
using Kalman filter. This algorithm produces good estimates and also deals with noise very
well. Different variations of the algorithm are tested (e.g., Extended Kalman filter) in order to
achieve the most accurate results. The accuracy is tested with motion-tracking system in the
laboratory. Finally, the optimized model is implemented as a mobile application and subsequently
evaluated underwater.