基于GAN网络的孔洞填充方法  

Hole filling technique utilizing GAN architecture

在线阅读下载全文

作  者:杨通 牛磊 Yang Tong;Niu Lei(Faculty of Geomatics,Lanzhou Jiaotong University,Lanzhou 730070,China;National-Local Joint Engineering Research Center of Technologies and Applications for National Geographic State Monitoring,Lanzhou Jiaotong University,Lanzhou 730070,China)

机构地区:[1]兰州交通大学,兰州730070 [2]兰州交通大学地理国情监测技术应用国家地方联合工程研究中心,兰州730070

出  处:《工程勘察》2025年第5期48-54,共7页Geotechnical Investigation & Surveying

基  金:国家自然科学基金项目(42171401,41771433);甘肃省自然科学基金(20JR10RA247)。

摘  要:点云是一种广泛应用于三维场景表示和处理的数据结构。然而,在真实世界中获取完整且准确的点云数据仍具有挑战性。因此,点云填充技术得到广泛研究,旨在通过生成缺失的点云数据来提升点云的完整性。本文提出一种基于生成对抗网络(GAN)的点云填充方法,该方法利用生成器网络生成缺失区域的点云,并通过判别器网络对其真实性进行评估。实验结果表明,本文方法可处理各种大小和形状的点云孔洞,并生成无缝衔接的修复结果。Point clouds serve as a fundamental data structure extensively employed for representing and processing 3D environments.Nevertheless,acquiring complete and precise point cloud data in the real world remains a formidable challenge.Consequently,extensive research efforts have been dedicated to point cloud inpainting techniques,with the primary objective of enhancing the comprehensiveness of point clouds by generating missing data.This paper introduces an innovative point cloud inpainting method,harnessing the power of Generative Adversarial Networks(GAN).The proposed approach utilizes a generator network to intelligently synthesize point cloud data within regions of absence,subsequently subjecting it to scrutiny through a discriminator network to assess its authenticity.Notably,our method’s experimental outcomes demonstrate its remarkable ability to adeptly address point cloud voids of diverse sizes and shapes,culminating in the seamless integration of inpainting regions with the existing data.In conclusion,this research represents a significant stride towards overcoming the challenges associated with incomplete point cloud data,offering promising prospects for various applications reliant on accurate 3D scene representation and analysis.

关 键 词:生成对抗网络 点云分割 孔洞填充 

分 类 号:TP391[自动化与计算机技术—计算机应用技术]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象