Region based Single-Stage Interference Mitigation and Target Detection

Dubey A, Fuchs J, Madhavan V, Lübke M, Weigel R, Lurz F (2020)

Publication Type: Conference contribution, Conference Contribution

Publication year: 2020

Event location: Florence IT

DOI: 10.1109/RadarConf2043947.2020.9266434


The inherent smaller radar cross sections of vulnerable road users resulting in smaller signal-to-noise-ratios make an accurate detection of them somewhat challenging. Mutual radar interference in typical automotive scenarios further imposes the difficulty of a target detection by additionally raising the noise floor. The traditional signal processing pipeline consists of multiple but separate stages for interference detection, mitigation and target detection. In this paper, a convolutional neural network based autoencoder architecture is used to perform a combined single-stage target detection while generalizing over different interference noise. The proposed approach achieves significant improvement over state-of-the-art methods while preserving the instance of each target and is able to identify them uniquely in case of a partial occlusion or overlapping of multiple targets.

Authors with CRIS profile

Related research project(s)

How to cite


Dubey, A., Fuchs, J., Madhavan, V., Lübke, M., Weigel, R., & Lurz, F. (2020). Region based Single-Stage Interference Mitigation and Target Detection. In Proceedings of the IEEE Radar Conference. Florence, IT.


Dubey, Anand, et al. "Region based Single-Stage Interference Mitigation and Target Detection." Proceedings of the IEEE Radar Conference, Florence 2020.

BibTeX: Download