基于群智感知的高效三维场景数据收集  

CROWDSOURCING BASED EFFECIENT 3D SCENE DATA COLLECTION METHOD

作　　者：王浩[1] 王旭 王黎征[1] WANG Hao;WANG Xu;WANG Li-zheng(Department of Digital Media and Art Design,Nanyang Institute of Technology,Nanyang 473004,China;First Affiliated Hospital,Nanyang Medical College,Nanyang 473000,China)

机构地区：[1]南阳理工学院数字媒体与艺术设计学院,河南南阳473004 [2]南阳高等医学专科学校第一附属医院,河南南阳473000

出　　处：《南阳理工学院学报》2021年第4期53-57,共5页Journal of Nanyang Institute of Technology

基　　金：河南省科技攻关项目(212102210384);河南省高等学校重点科研项目(21A520031);南阳市科技攻关项目(KJGG007);南阳理工学院交叉学科项目(520019);南阳理工学院教育教学改革项目(NIT2020JY-077)。

摘　　要：随着智慧城市的发展,城市中的三维场景理解系统越来越受到学者们的关注。然而,构建该系统必不可少的环节就是收集三维场景数据,如何实现高效的数据收集,成为人们研究的热点。本文利用群智感知机制给收集数据的志愿者规划路径,提出基于遗传算法的多邻域任务搜索方案。该方案能够利用遗传算法实现快速任务分配方法,并能够交换邻域之间的任务,从而实现快速的任务分配和3D数据收集。通过大量试验验证,该方法比同类算法性能更优。With the development of smart city,urban 3D scene understanding system receives more and more attentions.However,the essential part constructing the system is collecting 3D scene data,with the research hot point of implementing efficient data collection.It has become a popular trend to collect data sets of 3D scene understanding system by using crowdsourcing mechanism,although different scholars have proposed corresponding solutions.In this paper,we use crowdsourcing mechanism to plan the path for the participants to collect data,and propose a task search scheme based on dense target and variable neighborhood.The scheme can achieve fast task allocation and task exchange between neighbors,thus completing fast task collection.Through extensive experiments,this method achieves better performance than similar algorithms.

关 键 词：群智感知 三维场景数据收集 密目算法 遗传因子 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]