Cluster Quantity Distinguished by Geometric Angle Measurement
Zhongyang Shen · 2020
As an unsupervised learning algorithm applied in artificial intelligence, clustering algorithm has long time existed the problem of clustering quantity and position recognition. Aiming at the problem of clustering quantity recognition and point classification, a new clustering quantity recognition algorithm based on geometric angle measurement (QGAM) is proposed. The algorithm measures the quantity of clusters by geometric angle scanning and determines the boundaries of clusters by radius density measurement. The algorithm finds temporary cluster by searching the strongest density direction and location, repeats the process for the remaining points until all temporary clusters are found, and then identifies the farthest temporary cluster as the right one. Then the state of the remaining points is reset to initial state, and the same process is repeated until all clusters are found. The results show that QGAM is an effective method to identify the quantity of clusters.