3D-BSNet:双边特征和相似度量的点云实例分割网络  

3D-BSNet:bilateral features-similarity measure network for 3D instance segmentation

作　　者：田枫[1] 徐昕 刘芳[1] 刘宗堡[2] Tian Feng;Xu Xin;Liu Fang;Liu Zongbao(College of Computer&Information Technology,Northeast Petroleum University,Daqing Heilongjiang 163318,China;College of Geoscience,Northeast Petroleum University,Daqing Heilongjiang 163318,China)

机构地区：[1]东北石油大学计算机与信息技术学院,黑龙江大庆163318 [2]东北石油大学地球科学学院,黑龙江大庆163318

出　　处：《计算机应用研究》2023年第8期2514-2519,共6页Application Research of Computers

基　　金：黑龙江省自然科学基金资助项目(LH2021F004);黑龙江省哲学社会科学基金资助项目(22EDE389);黑龙江省高等学校教改工程项目(SJGZ20200037);东北石油大学研究生教育创新工程项目(JYCX_11_2020);东北石油大学引导性创新基金资助项目(2020YDL-11);黑龙江省省属本科高校基本科研业务费资助项目(KYCXTD201903);黑龙江省教育科学规划重点课题资助项目(GJB1421113)。

摘　　要：针对现有的三维分割方法在挖掘点云特征时,会忽略几何特征有效利用的问题,提出双边特征和相似度量的点云实例分割网络3D-BSNet(3D-bilateral feature and similarity measure network)。该网络主要由双边特征学习和轻量级相似度量两部分组成。首先提出一种包含基于子流形稀疏卷积的3D-UNet和多层感知机的双边特征提取模块,用于提取经过体素化处理的点云数据的语义特征和几何特征;然后设计一种结合通道维度和空间维度的双边注意力机制,用于减少双边特征聚合过程中产生的信息损失;最后开发一种轻量级相似度量模块,获取高维嵌入特征空间中邻近点云之间的相似性,并生成细粒度实例分割结果。实验表明,3D-BSNet在S3DIS和Scannet(v2)数据集上的多指标综合表现优越,其中在Scannet(v2)上的平均精确率比SSTNet提高了3.3%,有效提高了室内场景三维实例分割的精度。In order to solve the problem that the current three-dimensional segmentation methods ignored the effective use of geometric features when mining point cloud features,this paper proposed the point cloud instance segmentation network 3D-BSNet with bilateral features and similarity measure.The network mainly consisted of two parts,such as bilateral feature learning and lightweight similarity measurement.Firstly,this paper proposed the bilateral feature extraction module with 3D-UNet based on submanifold sparse convolution and multi-layer perceptron.This module could extract semantic and geometric features of point cloud data after voxelization.Secondly,this paper designed the bilateral attention mechanism combining channel dimension and spatial dimension.This module could reduce the information loss in the process of bilateral feature aggregation.Finally,this paper developed the lightweight similarity measurement module.This module could obtain the similarity between neighboring point clouds in the high-dimensional embedded feature space and even generate the segmentation results of fine-grained instances.Experiments show that 3D-BSNet performs well on S3DIS and Scannet(v2)dataset,and the average accuracy rate on Scannet(v2)is 3.3%higher than that of SSTNet,which effectively improves the precision of 3D instance segmentation.

关 键 词：点云实例分割 双边特征学习 双边注意力机制 轻量级相似度量 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]