【作者】 李丽娟； 刘爽； 林雪竹； 徐靓；

【Author】 LI Lijuan LIU Shuang LIN Xuezhu XU Liang (College of Opto-Electronic Engineering,Changchun University of Science and Technology,ChangChun,130022)

【机构】 长春理工大学光电工程学院；

【摘要】 为提高点云去噪精度和效率,提出一种基于BP神经网络相关算法,利用神经网络强大的非线性逼近能力,模拟真实曲面,将点云数据预测输出值与期望输出值对比,对超出要求误差范围的数据点进行删除,实现点云去噪。同时结合Matlab的计算功能和VB在图形界面开发方面的优势,生成脱离Matlab环境的点云去噪程序。为验证该去噪程序,本文对MK82导弹数据点进行实验,结果表明此程序对缩短点云预处理时间,提高逆向曲面重构精度,具有一定的工程实用价值和意义。更多还原

【Abstract】 In order to improve the accuracy and efficiency of the point cloud elimination,proposing a relative talgorithm of Back Propagation neural network,through BP network’s strong ability of non-linear approximation,real surface is simulated,and measured data of point clouds compared with predict output ones,deleting the data points that excessing requirement error range,realizing the point clouds elimination.Combined with computing capabilities of Matlab and advantages of VB in the graphical interface development,generating the process of point cloud denoising without Matlab environment.To prove the process of denoising,the experiment is carried on MK82 missile data points in this paper,the result indicates that the process has great practical value and significance on reducing point cloud pretreatment time and improving accuracy of reverse surface reconstruction.更多还原