Radar Classification for Traffic Intersection Surveillance based on Micro-Doppler Signatures

摘要

We propose a method for classifying objects within an intersection based on radar measurements. The complete processing chain beginning from raw data acquisition until target classification is elaborated. The range-Doppler processing for target detection, density-based spatial clustering of applications with noise (DBSCAN) clustering for associating the detections and a Kalman-filter based tracker for the multiple target scenario are implemented. As input for the classifier, features based on the micro-Doppler signatures were extracted and in a first step pedestrian and vehicles were discriminated by a support vector machine (SVM) classifier showing promising results from 300 recorded instances.