Development of an Integrated OCR System and a Digital Interface Twin for the NH-90 NTH Sea Lion

Maritime Search and Rescue (SAR) missions require operators to process large amounts of
information under demanding conditions. In the NH-90 NTH Sea Lion, relevant information
is presented through the Electro Optical System (EOS) interface, while direct access to the
underlying interface data is not available within the constraints of this work. This thesis
therefore investigates a non-invasive method for extracting EOS interface information and
reproducing the interface dynamically within a simulation environment.
First, the individual EOS interface elements are identified and analysed using video
material, literature research, and expert input. Based on these findings, a specialised
Optical Character Recognition (OCR) system is developed to extract textual and numerical
parameters from recorded video streams and store them as structured data. By exploiting
the fixed interface layout and limited character variability, the developed system achieves
an accuracy of 0.9998 on the labelled character dataset, with an estimated overall accuracy
between 0.9898 and 0.9998. Runtime measurements on the development system and a
cross-platform benchmark estimate suggest that real-time processing on the intended target
hardware is feasible, although direct validation remains necessary.
In addition, a functional and modular representation of the EOS interface is implemented
in the Unity-based PosAIdon simulation environment, providing a dynamic basis for future
simulation-based research.