刘如飞,柴永宁,朱健,俞家勇

• 1:山东科技大学测绘科学与工程学院

摘要(Abstract)：

针对三维激光点云数据生成360°全景深度图像存在像素分辨率不均匀的问题,提出一种顾及目标量测精度及可见度的全景深度图像生成方法,在保证全景影像表达地物的完整性的前提下提高其数据的存取精度。通过坐标转换和投影变换生成与全景影像匹配的全景深度图像;基于摄影成像原理分析摄影中心高度、深度值和像素分辨率之间的关系,得到不同深度处目标分辨率随摄影中心升高趋于一致的结论;综合分析地面目标分辨率和杆状目标尤其是树冠对树干的遮挡问题,确定特定场景下生成全景深度图像的最佳摄影中心位置并重新生成深度图像。实验分析表明,该方法能够在保证杆目标可见度的前提下提高地面目标量测精度。

关键词(KeyWords)： 三维激光点云;全景深度图像;成像原理;分辨率;测量精度

Abstract:

Keywords:

基金项目(Foundation): 国家重点研发计划项目（2016YFB0501705）;; 国家基础测绘科技计划项目（2016KJ0101）;; 山东省自然科学基金资助项目（ZR2019BD033）

作者(Author): 刘如飞,柴永宁,朱健,俞家勇

DOI: 10.16251/j.cnki.1009-2307.2021.01.023

参考文献(References)：

• [1] 李可宏,姜灵敏,龚永义.二维至三维图像/视频转换的深度图像提取方法综述[J].中国图象图形学报,2018,19(10):1393-1406.(LI Kehong,JIANG Lingmin,GONG Yongyi.Overview of depth image extraction methods for 2D to 3D image/video conversion[J].Chinese Journal of Image Graphics,2018,19 (10):1393-1406.)
• [2] 刘杰平,周华盛,余朗衡,等.基于视觉词典的深度图像生成算法[J].光学学报,2018,38(9):276-284.(LIU Jieping,ZHOU Huasheng,YU Langheng,et al.Depth image generation algorithm based on visual dictionary[J].Journal of Optics,2018,38 (9):276-284.)
• [3] WU Jiajun,ZHANG Chengkai,ZHANG Xiuming,et al.Learning shape priors for single-view 3D completion and reconstruction[J].2018,11215(10):673-691.
• [4] KUZNIETSOV Y,STüCKLER,JRG,LEIBE B.Semi-supervised deep learning for monocular depth map prediction[Z].[S.L.:s.n.],2017,1:2215-2223.
• [5] YAN Tao,ZHANG Fan,MAO Yiming,et al.Depth Estimation from a light field image pair with a generative model[J].IEEE Access,2019,7:12768-12778.
• [6] JANG Jang,PAIK Joonki.Dense depth map generation using a single camera with hybrid auto-focusing[C]//IEEE International Conference on Consumer Electronics-berlin.[S.l.]:IEEE,2016:277-278.
• [7] RAVANELLI R,NASCETTI A,CRESPI M.Kinect v2 and RGB stereo cameras integration for depth map enhancement[J].ISPRS - International Archives of the Photogrammetry,Remote Sensing and Spatial Information Sciences,2016,XLI-B5:699-702.
• [8] 卢秀山,俞家勇,田茂义,等.车载激光点云与序列化全景影像融合方法[J].中国激光,2018,497(5):245-252.(LU Xiushan,YU Jiayong,TIAN Maoyi,et al.Vehicle-borne laser point cloud and serialized panoramic image fusion method[J].China Laser,2018,497 (5):245-252.)
• [9] VARGA R,COSTEA A,HORA?IU F,et al.Super-sensor for 360-degree Environment Perception:Point Cloud Segmentation Using Image Features[C]//IEEE International Conference on Intelligent Transportation Systems.[S.l.]:IEEE,2017.
• [10] ZHANG Zhiqiang,ZHANG Xinchang,SUN Ying,et al.Road centerline extraction from very-high-resolution aerial image and LiDAR data based on road connectivity[J].Remote Sensing,2018,10(8):1-20.
• [11] LIU Wei,MA Lliyan,QIU Bo,et al.An efficient depth map preprocessing method based on structure-aided domain transform smoothing for 3D view generation[J].PLOS ONE,2017,12(4):1-20.
• [12] BIASUTTI P,AUJOL J F,BRéDIFM,et al.Range-image:incorporating sensor topology for LiDAR point cloud processing[J].Photogrammetric Engineering & Remote Sensing,2018,84(6):367-375.
• [13] HE Yuhang,CHEN Long,CHEN Jianda,et al.A novel way to organize 3D LiDAR point cloud as 2D depth map height map and surface normal map[C]//IEEE International Conference on Robotics & Biomimetics.[S.l.]:IEEE,2016:156-163.
• [14] 贾凯华,李和志,朱鹏.一种基于投影变换的深度图像生成算法[J].测绘工程,2016,25(12):61-64.(JIA Kaihua,LI Hezhi,ZHU Peng.A depth image generation algorithm based on projection transformation[J].Surveying and Mapping Engineering,2016,25 (12):61-64.)
• [15] ZHONG Ruofei,WEI Jiangxia,SU Wei,et al.A method for extracting trees from vehicle-borne laser scanning data[J].Mathematical and Computer Modelling,2013,58(3-4):733-742.