基于噪声分类的双边滤波点云去噪算法.docx

基于噪声分类的双边滤波点云去噪算法摘要：针对三维点云数据模型在去噪光顺中存在不同尺度噪声的问题提出一种基于噪声分类的双边滤波点云去噪算法。该算法首先将噪声细分为大尺度和小尺度噪声并使用统计滤波结合半径滤波对大尺度噪声进行去除；然后对三维点云数据进行曲率估计并对现有点云双边滤波进行改进增强其鲁棒性和保特征性；最后使用改进的双边滤波对小尺度噪声进行光顺实现三维点云数据模型的去噪、光顺。与单独使用双边滤波、Fleishman双边滤波相比改进算法在三维点云数据模型光顺平均误差指标上分别降低了50.53%和21.67%。实验结果表明该改进算法对噪声进行尺度的细分既提高了计算效率又避免了过光顺和细节失真较好地保持模型中的几何特征。关键词：点云去噪；双边滤波；统计滤波；半径滤波；尖锐特征中图分类号：TP391.41文献标志码：ADenoisingalgorithmforbilateralfilteredpointcloudbasedonnoiseclassificationYUANHuaPANGJiankeng*MOJianwenSchoolofInformationandCommunicationGuilinUniversityofElectronicTechnologyGuilinGuangxi541004ChinaAbstract：Focusingontheissuethatdifferentscalenoiseexistsindenoisingandsmoothingof3Dpointclouddatamodelabilateralfilteringdenoisingalgorithmfor3Dpointcloudbasedonnoiseclassificationwasproposed.Firstlythenoisepointsweresubdividedintothelargescaleandthesmallscalenoiseandthelargescalenoisewasremovedwithstatisticalfilteringandradiusfiltering.Secondlythecurvatureofthe3Dpointclouddatawasestimatedandthebilateralfilterwasimprovedtoenhancetherobustnessandsecurity.Finallythesmallscalenoisewassmoothedwiththeimprovedbilateralfiltertoachievethesmoothinganddenoisingof3Dpointclouds.ComparedwiththealgorithmssimplybasedonbilateralfilteringorFleishmanbilateralfilteringthesmoothingaverageerrorindexof3Dpointclouddatamodelobtainedbytheproposedmethodrespectivelydecreasedby50.53%and21.67%.Theexperimentalresultsshowthattheproposedalgorithmincreasestheefficiencyofcalculationbyscalesubdivionofnoisepointsandavoidsexcessivesmoothinganddetaildistortionwhichcanbettermaintainthegeometriccharacteristicsofthemodel.英文关键词Keywords：pointclouddenoising；bilateralfiltering；statisticalfiltering；radiusfiltering；sharpfeature0引言随着三维激光扫描技术的高速发展人们可以很容易获取真实世界中物体的表面三维点云数据因此三维点云数