Augmentation von Webcam-Bildern mit AIS-Daten
In this master's thesis a prototype was developed, which is able to show data from the Automatic Identification System (AIS) in an Augmented Reality (AR). Therefor Webcam-Images of a maritime scene where fused with AIS-Data, so that the user could see the information of the ship immediately in the image. Goal of this work was to exactly mark the ships with an AIS-Signal in the Webcam-Image. Even when the camera was moved the software should work fine. The type of the camera shouldn't influence the functionality of the software. A fluent transfer via the internet will be guaranteed. The ships without an AIS-Signal shouldn't be labeled. Disturbing and irrelevant objects where ignored. Only ships in the image space where labeled. Label shouldn't interleave each other. With increasing distance of the objects to the camera the label will get paler. The speed of the ship will influence the color of the label. In the label the name of the ship will be printed. The label will be placed in the image where the ship is. In this thesis the minimum requirements of the webcam where mentioned. The necessary configuration parameters where described in this thesis. Initially four concepts where presented and evaluated. During the project the concepts "location determination via transformation" and "ship detection with machine vision" were implemented. For the realization, the manipulation and the processing of image data is essentially. After an image preprocessing (rectifying of images) a transformation of ship positions from the WGS84-Format into pixel was implemented. The mathematical coherence and the usage of the open-source-bibliophile, OpenCV and cURL, were described in this thesis. After a preliminary result evaluation timing differences between image and AIS-Data were evaluated as the source of error. Through object detection this error was minimized. The given requirements could largely be complied. A detailed evaluation is part of this thesis.
München, Univ. der Bundeswehr, Master Thesis, 2017