[1]陈国军,孔李燕,张清伟,等.基于RGB-D三维点云目标分割[J].计算机技术与发展,2018,28(12):38-42.[doi:10．3969/j． issn．1673－629X．2018．12．008]
　CHEN Guojun,KONG Liyan,ZHANG Qingwei,et al.3D Point Cloud Target Segmentation Based on RGB－D Data[J].,2018,28(12):38-42.[doi:10．3969/j． issn．1673－629X．2018．12．008]
点击复制

基于RGB-D三维点云目标分割()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
28
期数:
2018年12期
页码:
38-42
栏目:
智能、算法、系统工程
出版日期:
2018-12-10

文章信息/Info

Title:
3D Point Cloud Target Segmentation Based on RGB－D Data
文章编号:
1673－629X(2018)12－0038－05
作者:
陈国军;孔李燕;张清伟;杨静;
中国石油大学(华东)计算机与通信工程学院;
Author(s):
CHEN Guo－jun; KONG Li－yan; ZHANG Qing－wei; YANG Jing
School of Computer ＆ Communication Engineering，China University of Petroleum，Qingdao 266580，China
关键词:
Kinect; 点云分割; 背景分割; 图像分割; Grab Cut; 法向量
Keywords:
Kinect; point cloud segmentation; background segmentation; image segmentation; Grab Cut; normal vector
分类号:
TP391.9
DOI:
10．3969/j． issn．1673－629X．2018．12．008
摘要:
三维点云的分割与分类是点云处理的关键步骤。针对点云模型分割出现的过分割和欠分割等分割不精确问题,提出一种基于RGB-D的背景点云目标分割方法,以提高点云模型的分割精度。利用Kinect相机对物体进行旋转拍摄可得到物体两帧背景点云和各角度的点云数据。算法利用背景帧根据深度信息对点云模型进行背景分割得到前景物体。结合图像分割和点云分割,利用Grab Cut算法对背景图像进行图像分割得到目标的RGB数据,随后对点云模型比较给定范围内的点的颜色信息和法向量进行点云数据的分割与合并,最后得到目标点云。实验结果表明,背景分割可以有效分割深度值小于背景的前景,结合图像分割有效地避免了过分割和欠分割问题。
Abstract:
The segmentation and classification of point clouds is a key step in point cloud processing． Aiming at the inaccurate segmenta- tion such as over segmentation and under segmentation，we propose a background point cloud target segmentation method based on RGB－ D to increase the segmentation accuracy of point cloud model． The two frame background point cloud and point cloud data of each angle can be obtained by using a Kinect camera． The algorithm uses the background frame to get the foreground object from the point cloud model based on the depth information． Then，combining image segmentation and point cloud segmentation，the Grab Cut algorithm is uti- lized to obtain the RGB data of the target． The target point cloud can be finally obtained by comparing RGB and normal vector of point cloud in the given range． Experiment shows that background segmentation can effectively segment the foreground of the depth value less than the background． And the problem of over segmentation and under segmentation is effectively avoided by combining image segmen- tation．

相似文献/References:

[1]黄露丹,严利民.基于Kinect深度数据的人物检测[J].计算机技术与发展,2013,(04):119.
　HUANG Lu-dan,YAN Li-min.Human Detection Based on Depth Data Acquired by Kinect[J].,2013,(12):119.
[2]谢亮,廖宏建,杨玉宝.基于Kinect的姿势识别与应用研究[J].计算机技术与发展,2013,(05):258.
　XIE Liang,LIAO Hong-jian,YANG Yu-bao.Recognition and Application Research of Kinect-based Gesture[J].,2013,(12):258.
[3]苏 卓,喻春阳.基于2D 图像变换的虚拟试衣算法[J].计算机技术与发展,2018,28(02):24.[doi:10．3969/j．issn．1673－629X．2018．02．006]
　SU Zhuo,YU Chunyang.A Virtual Fitting Algorithm Based on 2D Image Transformation[J].,2018,28(12):24.[doi:10．3969/j．issn．1673－629X．2018．02．006]
[4]杨雷坤[] [],李绪志[],王红飞[]. 基于手势识别技术的遥操作系统的设计与实现[J].计算机技术与发展,2015,25(09):1.
　YANG Lei-kun[] [],LI Xu-zhi[],WANG Hong-fei[]. Design and Implementation of Tele-operations System Based on Gesture Recognition Technology[J].,2015,25(12):1.
[5]杨张振.基于Kinect 的摔倒行为研究[J].计算机技术与发展,2018,28(04):179.[doi:10.3969/ j. issn.1673-629X.2018.04.038]
　YANG Zhang-zhen.Research on Fall Detection Based on Kinect[J].,2018,28(12):179.[doi:10.3969/ j. issn.1673-629X.2018.04.038]
[6]王劲东,武频. 一种基于Kinect的指尖检测算法[J].计算机技术与发展,2016,26(07):14.
　WANG Jin-dong,WU Pin. An Algorithm of Fingertip Detection Based on Kinect[J].,2016,26(12):14.
[7]罗章[][][],谭海波[][],李晓风[][][],等. Kinect运动捕获技术在健康医疗中的应用研究[J].计算机技术与发展,2016,26(08):104.
　LUO Zhang[][][],TAN Hai-bo[][],LI Xiao-feng[][][],et al. Research on Application of Kinect Motion Capture Technology in Health Care[J].,2016,26(12):104.
[8]耿曼[][],钱国明[],王宇[]. 虚拟试衣系统服装图像匹配算法研究[J].计算机技术与发展,2017,27(01):126.
　GENG Man[] [],QIAN Guo-ming[],WANG Yu[]. Research on Clothing Image Matching of Virtual Fitting Room[J].,2017,27(12):126.
[9]陈国军,张清伟,李开悦,等.基于RGB—D树状结构物体三维重建[J].计算机技术与发展,2018,28(12):142.[doi:10．3969／j．issn．1673—629X．2018．12．030]
　CHENGuo-jan,ZHANG Qing-wei,LI Kaiyue,et al.3D Reconstruction of Tree Structure Based on RGB—D[J].,2018,28(12):142.[doi:10．3969／j．issn．1673—629X．2018．12．030]
[10]李鑫,陈建新,陈克坚,等.基于Kinect的体育运动自训练系统[J].计算机技术与发展,2019,29(04):122.[doi:10. 3969 / j. issn. 1673-629X. 2019. 04. 025]
　LI Xin,CHEN Jian-xin,CHEN Ke-jian,et al.Kinect-based Sports Self-training System[J].,2019,29(12):122.[doi:10. 3969 / j. issn. 1673-629X. 2019. 04. 025]

更新日期/Last Update: 2018-12-10