The DBSCAN clustering algorithm sets the class to the largest collection of density-connected objects, which can group very high density parts into one class, and can find clusters of classes of all possible shapes in the target object where "noise" exists.
正在翻译中..