摘要
空洞卷积胶囊网络环境中,激光雷达点云数据分类处理方法在数据采样后存在数据分辨率下降、数据丢失的问题,导致最终分类结果较差。为此,研究空洞卷积胶囊网络环境中的激光雷达点云数据分类处理方法。对激光雷达点云数据进行空洞卷积预处理,在分析其光谱特征的条件下执行分类指令,进一步划分分类信息,获取清晰完整的分类数据,由此增强整体分类效果。根据分析的实验结果可知,该处理方法的错误分类面积较小,分类精度较高,总体分类精度为99.67%,这表明该方法能够有效实现激光雷达点云数据分类处理。
In the hollow convolutional capsule network environment,the data classification processing method of lidar point cloud has the problems of data resolution degradation and data loss after data sampling,resulting in poor final classification results.Therefore,the classification processing method of lidar point cloud data in the cavity convolutional capsule network environment is studied.The hole convolution preprocessing of lidar point cloud data is carried out,and the classification instruction is executed under the condition of analyzing its spectral characteristics,and the classification information is further divided to obtain clear and complete classification data,thereby enhancing the overall classification effect.According to the experimental results of the analysis,the error classification area of the processing method is small,the classification accuracy is high,and the overall classification accuracy is 99.67%,which indicates that the method can effectively realize the classification processing of lidar point cloud data.
作者
陈雨
陈亮
CHEN Yu;CHEN Liang(RCG Geosystems(Beijing)Co.,Ltd.,Beijing 100125,China)
出处
《电子设计工程》
2024年第22期31-36,共6页
Electronic Design Engineering
关键词
空洞卷积技术
卷积胶囊网络环境
激光雷达
点云数据
数据分类
cavity convolution technology
convolutional capsule network environment
lidar
point cloud data
data classification
