DBSCAN and K-Means Hybrid Clustering Based Automatic Dental Feature Detection

According to the characteristics of dental meshes, this paper proposed an automatic dental feature detection algorithm based on DBSCAN（density-based spatial clustering of applications with noise） and K-Means hybrid clustering. First, average curvatures and geodesic distances were used to better represent the distances between partition. Second, DBSCAN was applied to projective points which have larger Z values in order to obtain the number of clusters and center points as the input of next step. Next, we used K-Means algorithm to generate partitions of the dental mesh. Finally, the feature detection algorithm was employed to obtain feature points on each partition. Experiments show that this algorithm can accurately detect dental features. Compared to the previous automatic algorithm, the accuracy is significantly improved and the operations are also simplified.