基于局部结构模型库的建筑物LiDAR点云去噪

Local structure model library based denoising method for building roof data from airborne LiDAR

通过研究建筑物结构中点、线基本结构模型,构建了相应的基本结构模型库。为了验证该模型库的有效性,利用模型库产生的仿真数据构造训练数据实现了一种在K-SVD框架下的建筑物屋顶LiDAR点云数据去噪方法。实验结果表明,所构建的建筑物模型库对建筑物点云数据具有非常好的表示能力,能够有效地应用于点云数据的去噪。

For consideration that no matter the building is complex or not, it is composed by basic structures such as point/corner, edge/line, or plane, a local structure model library for both corner and linear structures is constructed. To evaluate the performance of the constructed library, simulated data generated according to the local structure library is used as training data for K-SVD denoising method to denoise building roofs data from airborne LiDAR. Experimental re- suits illustrate that the representation capability of the proposed local structure library is very well.