Figure 1.

Running scenario of the DP clustering stage using Dataset C1_005. (a) Data points projected in a two-dimensional feature subspace after LDA. This subspace is considered to be optimal for the within-class scatters are relatively small and intra-class scatters are relatively large. (b) Schematic diagram with the local density $\rho$ and the minimum distance $\delta$ as the horizontal and vertical axes. The points circled in three colors are the screened cluster centers. (c) Scatter plot of 3 clusters obtained by DP. The three colors of red, green and blue respectively represent the clustering results of cluster 1, cluster 2 and cluster 3. The black dots represent the cluster centers. (d) The waveforms of three cluster centers.