For many radar applications, a grouping of radar reflections that belong to the same object is needed. Unsupervised clustering algorithms are commonly used for this task. However, the number and density of measured reflections of an object depends on various parameters and therefore unsupervised algorithms often fail to identify all points that should be part of the same cluster. We propose a method to incorporate learned knowledge about the data into the clustering algorithm and show that this new method outperforms unsupervised approaches.
Supervised Clustering for Radar Applications: On the Way to Radar Instance Segmentation
2018-04-01
2993044 byte
Conference paper
Electronic Resource
English