摘要
针对复杂场景中三维点云噪声数据影响人体三维重建效率、精度的问题,提出一种基于深度信息的三维点云去噪方法。首先使用Kinect三维扫描仪获取场景中人体的三维深度噪声数据,采用多模型数据融合方法进行预处理,然后通过改进的C均值算法去除离群点噪声并进行聚类,最后利用深度数据双边滤波方法对高频信息进行处理。通过实验对比验证,论文方法能够有效地除去大尺度噪声和光顺小尺度噪声,并保留数据的完整性。
The point cloud data noise affects the body 3d reconstruction efficiency and accuracy in complex scene. To solve the problems,a method of denoising based on the kinect depth information of 3d point cloud is presented. First of all,by using the kinect device scan scene,the 3d noise data of body is obtained. A method of multiple model data fusion is used for data preprocessing,then,by the improved c-means algorithm,this noise is removed and dustered. Finally,by using the depth data of bilateral filtering method,the high frequency information is presented. By comparison with the experimental verification,the method can effectively remove the large scale suitable small noise,noise and light and maintain the integrity of the data.
作者
杜棋东
DU Qidong(Educational Technology Center of Guangzhou Railway Polytechnic,Guangzhou 510430)
出处
《计算机与数字工程》
2018年第11期2170-2174,共5页
Computer & Digital Engineering
基金
国家自然科学基金青年科学基金项目(编号:61503143)
广东省科技计划项目(编号:2015A030401005)
2016年广州铁路职业技术学院校级课题(编号:GTXYY1614)资助
