EE Seminar: Ship Detection Method Based on Images from a Geostationary Satellite
https://zoom.us/j/97902155480?pwd=zHu2G6zd0UgkGkWsyb7ZIxvYCubGsi.1
Meeting ID: 979 0215 5480
Passcode: 9NurbW
Electrical Engineering Systems ZOOM Seminar
Speaker: Lior Yannai
M.Sc. student under the supervision of Prof. Ofer Amrani
Wednesday, 18th March 2026, at 12:00
Ship Detection Method Based on Images from a Geostationary Satellite
Abstract
In recent years, the space industry has undergone a paradigm shift, marked by the increasing involvement of private-sector companies such as SpaceX and Blue Origin. These companies have driven technological advancements in launch systems, notably through reusable rockets that significantly reduce costs. This evolution has made the deployment of small and nano-satellites more feasible and widespread, leading to large-scale satellite constellations for applications such as global communications and Earth observation. One example is Tel Aviv University’s TAU-SAT1 to TAU-SAT3 series, which highlight the growing accessibility of space-based experimentation.
This study focuses on the detection of maritime vessels using low-cost geostationary satellite platforms, with high temporal resolution and low spatial resolution of approximately 200 meters per pixel. In this research, an algorithm was developed for ship detection under those conditions, where the algorithm was based on an online database that was resolution-adjusted.
The study assumes that the imagery will be captured over a fixed area, so that at each point in time a similar image of the same region will be obtained, with changes only in the positions of moving ships. Another assumption is that in a very small area, at most a single ship will be present at any given time, which will need to be detected and tracked.
The broader objective is to demonstrate that useful maritime surveillance can be achieved even with significantly constrained imaging resources. This is especially relevant for future low-cost nano-satellite missions, where trade-offs between resolution, cost, and onboard processing capabilities are critical. By building and evaluating a simplified yet robust detection framework, this study aims to contribute to more accessible and scalable space-based maritime monitoring solutions.
This study confronts the challenge of performance evaluation in the absence of publicly available datasets that match the severely degraded imaging conditions under consideration. Although several existing repositories were investigated, most proved unsuitable for our objectives.
Ultimately, after identifying a dataset with an enough ship imagery and contextual variety (but without the temporal requirements), we carried out a quantitative assessment of False-Alarm and Miss-Detection rates across a range of vessel sizes and environmental scenarios. The results confirm that reliable detection is attainable, but only when the system is designed with careful attention to the inherent constraints of low-resolution, high-noise satellite data.
-סמינר זה ייחשב כסמינר שמיעה לתלמידי תואר שני ושלישי-
This Seminar Is Considered A Hearing Seminar For Msc/Phd Students-
כדי לקבל קרדיט שמיעה יש לחתום שם מלא ומספר ת.ז. בצ'ט
