基于坡度滤波的无人机测绘点云特征提取与滤波分类研究

Research on Feature Extraction and Filtering Classification of Unmanned Aerial Vehicle Surveying Point Clouds Based on Slope Filtering

摘要:

为提高无人机测绘点云数据的质量与利用率,利用坡度滤波技术优化设计无人机测绘点云特征提取与滤波分类方法。利用硬件设备获取无人机测绘点云数据,通过粗配准和精准配准两个步骤,实现对初始点云数据的配准工作。通过对无人机测绘点云数据的滤波处理,降低初始数据中的干扰项。利用坡度滤波技术提取无人机测绘点云数据的地形、纹理、形状等特征,根据特征相似度的计算结果,完成测绘点云的滤波分类。通过性能测试实验得出结论:与传统方法相比,优化设计方法得出点云数据的信噪比提高41.22,特征提取占比所有提升、冗余度得到明显降低,分类查全率和查准率分别提高了1.25%和2.1%。

Abstract:

To improve the quality and utilization of unmanned aerial vehicle point cloud data, slope filtering technology is used to optimize the design of unmanned aerial vehicle point cloud feature extraction and filtering classification methods. Utilizing hardware devices to obtain drone surveying point cloud data, the initial point cloud data is registered through two steps: coarse registration and precise registration. By filtering and processing the point cloud data measured by drones, the interference terms in the initial data are reduced. Using slope filtering technology to extract terrain, texture, shape and other features from unmanned aerial vehicle (UAV) point cloud data, and based on the calculation of feature similarity, complete the filtering and classification of the surveyed point cloud. Through performance testing experiments, it was concluded that compared with traditional methods, the optimized design method resulted in a 41.22% improvement in the signal-to-noise ratio of point cloud data, an overall increase in feature extraction, a significant reduction in redundancy, and a 1.25% and 2.1% increase in classification recall and precision, respectively.