Optimization and Simulation of Missing Data Classification Coefficients for 3D Point Cloud Based on DBSCAN
At present,low quality of 3D point cloud data may cause classification difficulty.Therefore,this article presented a method of optimizing 3D point cloud data classification based on DBSCAN algorithm.First,we prepro-cessed the 3D point cloud data and filled in the missing data,thus ensuring data integrity.Then,we eliminated invalid points far from the main body of the 3D point cloud by the passthrough filtering method.Meanwhile,we adopted the K-D tree and KNN algorithm to improve the statistical filtering and filter the outliers in 3D point cloud data,thus opti-mizing the quality of original 3D point cloud data.Moreover,we used a beetle swarm optimization algorithm to improve the DBSCAN algorithm and select two parameters,namely the neighborhood search radius of the DBSCAN algorithm and the minimum number of objects contained in the search neighborhood.Finally,we input the optimized 3D point cloud data into the improved DBSCAN algorithm,thus achieving the 3D point cloud data classification.Experimental results show that the C-H coefficient and profile coefficient of the proposed method are larger and the D-B coefficient is smaller.
DBSCAN algorithm3D point cloud dataData classificationData preprocessingBeetle Swarm Opti-mization Algorithm,BSO
NETL
NSTL
万方数据
