mmwave.clustering.clustering

mmwave.clustering.clustering.associate_clustering(new_cluster, pre_cluster, max_num_clusters, epsilon, v_factor, use_elevation=False)

Associate pre-existing clusters and the new clusters.

The function performs an association between the pre-existing clusters and the new clusters, with the intent that the cluster sizes are filtered.

Parameters:

• new_cluster –
• pre_cluster –
• max_num_clusters –
• epsilon –
• v_factor –
• use_elevation –

mmwave.clustering.clustering.radar_dbscan(det_obj_2d, weight, doppler_resolution, use_elevation=False)

DBSCAN for point cloud. Directly call the scikit-learn.dbscan with customized distance metric.

DBSCAN algorithm for clustering generated point cloud. It directly calls the dbscan from scikit-learn but with customized distance metric to combine the coordinates and weighted velocity information.

Parameters:

• det_obj_2d (ndarray) – Numpy array containing the rangeIdx, dopplerIdx, peakVal, xyz coordinates of each detected points. Can have extra SNR entry, not necessary and not used.
• weight (float) – Weight for velocity information in combined distance metric.
• doppler_resolution (float) – Granularity of the doppler measurements of the radar.
• use_elevation (bool) – Toggle to use elevation information for DBSCAN and output clusters.

Returns:

Numpy array containing the clusters’ information including number of points, center and

size of the clusters in x,y,z coordinates and average velocity. It is formulated as the structured array for numpy.

Return type:

clusters (np.ndarray)