Filter-based Segmentation of Automotive SAR Images

Hoffmann M, Noegel T, Schüßler C, Vossiek M, Schütz M, Gulden P (2022)

Publication Type: Conference contribution

Publication year: 2022

Event location: New York City

DOI: 10.1109/RadarConf2248738.2022.9764276


This paper presents an approach to the segmentation of automotive synthetic aperture radar (SAR) images generated with a state-of-the-art multiple-input multiple-output frequency modulated continuous wave radar. Due to the very high image resolution and large signal-to-noise ratio, it is possible to directly distinguish different free spaces, extended objects, or target areas, such as walls or cars, and point-like targets, including posts and trees. The latter are detected using a constant false alarm rate algorithm, whereas free spaces and target areas are segmented by using Otsu's method after morphologically processing the SAR image. With this simple, filter-based segmentation, the perception of the vehicle's environment can be improved without complex post-processing that is typically required in standard radar gridmaps.

Authors with CRIS profile

Additional Organisation(s)

Involved external institutions

How to cite


Hoffmann, M., Noegel, T., Schüßler, C., Vossiek, M., Schütz, M., & Gulden, P. (2022). Filter-based Segmentation of Automotive SAR Images. In IEEE (Eds.), Proceedings of the 2022 IEEE Radar Conference. New York City.


Hoffmann, Marcel, et al. "Filter-based Segmentation of Automotive SAR Images." Proceedings of the 2022 IEEE Radar Conference, New York City Ed. IEEE, 2022.

BibTeX: Download