[1]王爱兵,杨晓文,韩 燮,等.优化球查询算法的点云分割[J].计算机技术与发展,2022,32(08):55-59.[doi:10. 3969 / j. issn. 1673-629X. 2022. 08. 009]
　ANG Ai-bing,YANG Xiao-wen,HAN Xie,et al.Point Cloud Segmentation of Optimized Ball Query Algorithm[J].,2022,32(08):55-59.[doi:10. 3969 / j. issn. 1673-629X. 2022. 08. 009]
点击复制

优化球查询算法的点云分割()
分享到：

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

卷:
32
期数:
2022年08期
页码:
55-59
栏目:
图形与图像
出版日期:
2022-08-10

文章信息/Info

Title:
Point Cloud Segmentation of Optimized Ball Query Algorithm
文章编号:
1673-629X(2022)08-0055-05
作者:
王爱兵; 杨晓文; 韩 燮; 郭新东; 彭志斌; 郭子军; 贾彩琴
中北大学 大数据学院,山西 太原 030051
Author(s):
ANG Ai-bing; YANG Xiao-wen; HAN Xie; GUO Xin-dong; PENG Zhi-bin; GUO Zi-jun; JIA Cai-qin
School of Big Data,North University of China,Taiyuan 030051,China
关键词:
球查询; 多尺度; 多分辨率; PointNet++; K-近邻; 局部特征
Keywords:
ball query; multi-scale; multi-resolution; PointNet++; K-nearest neighbor; local features
分类号:
TP391. 41
DOI:
10. 3969 / j. issn. 1673-629X. 2022. 08. 009
摘要:
为丰富 PointNet++网络局部特征的表征能力、加强细节特征的表达效果、提高点云模型的分割精度,针对 PointNet++中多尺度和多分辨率算法都采用的点集重叠划分方法冥球查询算法进行了研究。 在 PointNet++中,球查询算法随机选取球形邻域内的特征点提取局部特征,导致局部特征表达效果欠佳。 为加强局部特征表征能力,引入 K-近邻优化策略,将球邻域内的特征点按照与中心点的距离由近及远排序。 在局部特征提取过程中,当球邻域内的点数超过需要的特征点时选取距中心点相对较近的一批点作为局部特征提取点;当球邻域中的点数少于需要的特征点时,选取距中心点最近的特征点复制多次,补齐特征表示。 将优化的球查询算法应用于 PointNet++分割网络,利用 S3DIS 和 ShapeNetPart 作为数据集验证算法的有效性。 实验结果表明,优化的球查询算法丰富了网络的局部特征表征能力,强化了细节特征的表达效果,提高了分割精度。
Abstract:
In order to enrich the representation ability of the local features of the PointNet++ network,strengthen the expression effect ofthe detailed features and improve the segmentation accuracy of point cloud models,we study the sphere query algorithm,which is thepoint set overlap division method used in the multi-scale and multi-resolution algorithms in PointNet++. In PointNet++,the ball queryalgorithm randomly selects feature points in the spherical neighborhood to extract local features, which leads to poor local featureexpression. In order to strengthen the local feature representation ability,we introduce the K-nearest neighbor optimization strategy,andthe feature points in the spherical neighborhood are sorted from near to far according to the distance from the center point. During localfeature extraction,when the number of points in the neighborhood of the ball exceeds the required feature points,select a group of pointsrelatively close to the center point as the local feature extraction points. When the number of points in the neighborhood of the ball is lessthan the required feature points,the feature points closest to the center point are selected and copied multiple times to complement thefeature representation. The optimized ball query algorithm is applied to the PointNet + + segmentation network, and S3DIS andShapeNetPart are used as a data set to verify the effectiveness of the algorithm. The experiment shows that the optimized ball queryalgorithm enriches the local feature representation ability of the network, strengthens the expression effect of detailed features, andimproves the segmentation accuracy.

相似文献/References:

[1]刘亚东 李翠华.基于多尺度边缘和局部熵原理的前方车辆检测[J].计算机技术与发展,2008,(03):200.
　LIU Ya-dong,LI Cui-hua.Preceding Vehicle Detection Based on Multiple Scale Edge and Local Entropy[J].,2008,(08):200.
[2]胡媛媛 蔡光程.基于多结构元多尺度的形态学边缘检测[J].计算机技术与发展,2008,(11):97.
　HU Yuan-yuan,CAI Guang-cheng.Morphological Edge Detection Based on Multi- Structuring Elements and Multi- Scale[J].,2008,(08):97.
[3]张伟,王军锋,王涛,等.一种基于改进算子的形态学边缘检测算法[J].计算机技术与发展,2013,(06):23.
　ZHANG Wei,WANG Jun-feng,WANG Tao,et al.An Improved Edge Detection Algorithm Based on Morphologic Operators[J].,2013,(08):23.
[4]于笃发,邵建华,张晶如.基于小波自适应阈值图像去噪方法的研究[J].计算机技术与发展,2013,(08):250.
　YU Du-fa,SHAO Jian-hua,ZHANG Jing-ru.Research on Image Denoising Based on Wavelet Adaptive Threshold[J].,2013,(08):250.
[5]李梦涵,熊淑华,熊文,等. 多尺度级联行人检测算法的研究与实现[J].计算机技术与发展,2014,24(08):10.
　LI Meng-han,XIONG Shu-hua,XIONG Wen,et al. Research and Realization of Pedestrian Detection Algorithm by Multi-scale Cascaded Features[J].,2014,24(08):10.
[6]高建[],杨刚[]. 基于FLST变换的多尺度面状地物提取方法[J].计算机技术与发展,2014,24(12):167.
　GAO Jian[],YANG Gang[]. Multi-scale Extraction Method of Area Feature Based on FLST[J].,2014,24(08):167.
[7]卢健,黄杰,潘峰. 基于多尺度各向异性高斯核的彩色图像边缘检测算法[J].计算机技术与发展,2016,26(05):66.
　LU Jian,HUANG Jie,PAN Feng. Color Image Edge Detection Algorithm Based on Multi-scale Anisotropic Gaussian Filter[J].,2016,26(08):66.
[8]苑金辉,乔 艳,费烨琳,等.基于深度迁移学习的心脏 MRI 图像左心室分割[J].计算机技术与发展,2021,31(06):35.[doi:10. 3969 / j. issn. 1673-629X. 2021. 06. 007]
　YUAN Jin-hui,QIAO Yan,FEI Ye-lin,et al.Left Ventricular Segmentation in Cardiac MRI Images Based onDeep Transfer Learning[J].,2021,31(08):35.[doi:10. 3969 / j. issn. 1673-629X. 2021. 06. 007]
[9]肖顺亮,强赞霞,刘卫光.基于 CSP 改进用于拥挤情况的行人检测算法[J].计算机技术与发展,2021,31(07):52.[doi:10. 3969 / j. issn. 1673-629X. 2021. 07. 009]
　XIAO Shun-liang,QIANG Zan-xia,LIU Wei-guang.An Improved Pedestrian Detection Algorithm for Crowd Based on CSP[J].,2021,31(08):52.[doi:10. 3969 / j. issn. 1673-629X. 2021. 07. 009]
[10]王 贺,王志宝,陈良富,等.基于多尺度复合金字塔模型的缓存策略研究[J].计算机技术与发展,2022,32(02):101.[doi:10. 3969 / j. issn. 1673-629X. 2022. 02. 016]
　WANG He,WANG Zhi-bao,CHEN Liang-fu,et al.Research on Caching Strategy Based on Multi-scale Compound Pyramid Model[J].,2022,32(08):101.[doi:10. 3969 / j. issn. 1673-629X. 2022. 02. 016]

更新日期/Last Update: 2022-08-10