单目多视角全景视觉感知三维重构技术研究
|
摘要
从图像中恢复出几何实体的三维形状是计算机视觉研究领域的一个核心课题.本文提出了一种基于单目全景成像的三维重构方法.通过单目多视角全景视觉传感器在一个摄像平面上同时获取从多个不同视角拍摄的被测物体图像,采用Otsu算法对图像进行分割,提取前景信息,用轮廓体素极坐标遍历得到几何实体的轮廓采样点数据,重构出几何实体的三维形状.单目多视角全景视觉传感器避免了通用多视角成像中摄像机的颜色系统以及内外参数难以保持一致的问题,减少了硬件成本和立体匹配的复杂度,增加了实时性;通过轮廓体素极坐标遍历得到物体的三维点云数据是一种明确、不含二义性的信息,具有较好的鲁棒性.通过实际系统应用验证,该方法是一种行之有效的三维重构方法.
To recover 3D geometry from images is a core research in computer vision and computer graphics. An algorithm of Objectcentered 3D reconstruction based monocular multi-view images is presented. Using monocular multi-view sensor to gain one panorama picture with multi-view images of geometric object,segmenting this picture by Otsu algorithm,bringing up foreground information,acquiring outline sample points of the object by using traversal method of outline voxels based on polar coordinates,then reconstruct 3D geometric modeling. Using monocular multi-view stereo sensor can overcome the problem of usual multi-view images that the inner and outer parameters and color system of different cameras is difficult to keep the same,can reduce complexity of multi-view matching calculation and cost of hardware,can increase real-time performance. 3D points of object reconstruction by outline voxel traversal of polar coordinates is clear and unambiguous,so it has good robustness. This method has been verified in the actual system.
To recover 3D geometry from images is a core research in computer vision and computer graphics. An algorithm of Objectcentered 3D reconstruction based monocular multi-view images is presented. Using monocular multi-view sensor to gain one panorama picture with multi-view images of geometric object,segmenting this picture by Otsu algorithm,bringing up foreground information,acquiring outline sample points of the object by using traversal method of outline voxels based on polar coordinates,then reconstruct 3D geometric modeling. Using monocular multi-view stereo sensor can overcome the problem of usual multi-view images that the inner and outer parameters and color system of different cameras is difficult to keep the same,can reduce complexity of multi-view matching calculation and cost of hardware,can increase real-time performance. 3D points of object reconstruction by outline voxel traversal of polar coordinates is clear and unambiguous,so it has good robustness. This method has been verified in the actual system.
引文
[1]Guo Wei-qing,Tang Yi-ping,Lu Shao-hui,et al.Review of 3Dstereo vision measure and reconstruction based on mirror image[J].Computer Science,2016,43(9):1-10.
[2]Tang Yi-ping.Sensing technologies and applications of internet of things[M].Beijing:Electronic Industry Press,2015.
[3]Grzegorzek M,Theobalt C,Koch R,et al.Time-of-flight and depth imaging.sensors,algorithms and applications:dagstuhl seminar2012 and GCPR workshop on imaging new modalities[J].Lecture Notes in Computer Science,2013,8200(2):354-360.
[4]Lipson H,Shpitalni M.Optimization-based reconstruction of a 3Dobject from a single freehand line drawing[C]//ACM SIGGRAPH2007 Courses,ACM,2007:45.
[5]Blanz V.Face recognition based on a 3D morphable model[C]//International Conference on Automatic Face and Gesture Recognition,IEEE,2006:617-624.
[6]Maghari A,Venkat I,Liao I Y,et al.Adaptive face modelling for reconstructing 3D face shapes from single 2D images[J].Computer Vision Iet,2014,8(5):441-454.
[7]Sugita K,Sekine M,Nishiyama M,et al.Image processing apparatus,image processing system,image processing method,and computer program product[J].And Then Becomes a Constant Value After Time,2018,1(1):65-70.
[8]Chen Y,Jin X,Dai Q.Distance estimation based on light field geometric modeling[C]//IEEE International Conference on Multimedia&Expo Workshops,IEEE Computer Society,2017:43-48.
[9]Saxena A,Sun M,Ng A Y.Make3d:learning 3d scene structure from a single still image[J].Pattern Analysis and Machine Intelligence,IEEE Transactions on,2009,31(5):824-840.
[10]Wang X,Hou C,Pu L,et al.A depth estimating method from a single image using FoE CRF[J].Multimedia Tools and Applications,2015,74(21):9491-9506.
[11]Li Zhang,Guillaume Dugas-Phocion,Jean-Sebastien Samson,et al.Single-view modelling of free-form scenes[J].Computer Animation and Virtual Worlds,2010,13(4):225-235.
[12]Su Bo,Qiu Xian-jie,Wang Zhao-qi.Review of geometric modeling technology based on image[J].Journal of Computer Research and Development,2010,47(3):549-560.
[13]Jiang B.Real-time multi-resolution edge detection with pattern analysis on graphics processing unit[J].Journal of Real-time Image Processing,2018,14(2):293-321.
[14]Hoiem D,Efros A A,Hebert M.Recovering surface layout from an image[J].International Journal of Computer Vision,2007,75(1):151-172.
[15]Tang Yi-ping,Yang Rui-da,Zhou Wei-min,et al.Based on monocular mult-view machine vision real shoe replication device and last manufacturing methods:CN104573180A[P].2015-04-29.
[16]Guo Wei-qing,Tang Yi-ping,Xia Shao-jie,et al.Research on 3Dreconstruction based on panorama image of specular reflection[J].High Technology Communication,2015,25(z2):878-885.
[17]Yang Z,Lu S,Wu T,et al.Detection of morphology defects in pipeline based on 3D active stereo omnidirectional vision sensor[J].Iet Image Processing,2018,12(4):588-595.
[18]Tang Yi-ping,Wu Ting,Yuan Gong-ping,et al.A 3D omni-directional vision sensor for detection of morphology defects in pipelines[J].Journal of Instrument and Meter,2017,38(3):726-733.
[1]郭伟青,汤一平,鲁少辉,等.基于镜面成像技术的三维立体视觉测量与重构综述[J].计算机科学,2016,43(9):1-10.
[2]汤一平.物联网感知技术与应用[M].北京:电子工业出版社,2015.
[12]束搏,邱显杰,王兆其.基于图像的几何建模技术综述[J].计算机研究与发展,2010,47(3):549-560.
[15]汤一平,杨瑞达,周伟敏,等.基于单目多视角机器视觉的真人鞋型复制装置和鞋楦制造方法:CN104573180A[P].2015-04-29.
[16]郭伟青,汤一平,夏少杰,等.基于镜面折反射全景成像的三维重建方法研究[J].高技术通讯,2015,25(z2):878-885.
[18]汤一平,吴挺,袁公萍,等.适用于管道内形貌检测的3D全景视觉传感器[J].仪器仪表学报,2017,38(3):726-733.
[2]Tang Yi-ping.Sensing technologies and applications of internet of things[M].Beijing:Electronic Industry Press,2015.
[3]Grzegorzek M,Theobalt C,Koch R,et al.Time-of-flight and depth imaging.sensors,algorithms and applications:dagstuhl seminar2012 and GCPR workshop on imaging new modalities[J].Lecture Notes in Computer Science,2013,8200(2):354-360.
[4]Lipson H,Shpitalni M.Optimization-based reconstruction of a 3Dobject from a single freehand line drawing[C]//ACM SIGGRAPH2007 Courses,ACM,2007:45.
[5]Blanz V.Face recognition based on a 3D morphable model[C]//International Conference on Automatic Face and Gesture Recognition,IEEE,2006:617-624.
[6]Maghari A,Venkat I,Liao I Y,et al.Adaptive face modelling for reconstructing 3D face shapes from single 2D images[J].Computer Vision Iet,2014,8(5):441-454.
[7]Sugita K,Sekine M,Nishiyama M,et al.Image processing apparatus,image processing system,image processing method,and computer program product[J].And Then Becomes a Constant Value After Time,2018,1(1):65-70.
[8]Chen Y,Jin X,Dai Q.Distance estimation based on light field geometric modeling[C]//IEEE International Conference on Multimedia&Expo Workshops,IEEE Computer Society,2017:43-48.
[9]Saxena A,Sun M,Ng A Y.Make3d:learning 3d scene structure from a single still image[J].Pattern Analysis and Machine Intelligence,IEEE Transactions on,2009,31(5):824-840.
[10]Wang X,Hou C,Pu L,et al.A depth estimating method from a single image using FoE CRF[J].Multimedia Tools and Applications,2015,74(21):9491-9506.
[11]Li Zhang,Guillaume Dugas-Phocion,Jean-Sebastien Samson,et al.Single-view modelling of free-form scenes[J].Computer Animation and Virtual Worlds,2010,13(4):225-235.
[12]Su Bo,Qiu Xian-jie,Wang Zhao-qi.Review of geometric modeling technology based on image[J].Journal of Computer Research and Development,2010,47(3):549-560.
[13]Jiang B.Real-time multi-resolution edge detection with pattern analysis on graphics processing unit[J].Journal of Real-time Image Processing,2018,14(2):293-321.
[14]Hoiem D,Efros A A,Hebert M.Recovering surface layout from an image[J].International Journal of Computer Vision,2007,75(1):151-172.
[15]Tang Yi-ping,Yang Rui-da,Zhou Wei-min,et al.Based on monocular mult-view machine vision real shoe replication device and last manufacturing methods:CN104573180A[P].2015-04-29.
[16]Guo Wei-qing,Tang Yi-ping,Xia Shao-jie,et al.Research on 3Dreconstruction based on panorama image of specular reflection[J].High Technology Communication,2015,25(z2):878-885.
[17]Yang Z,Lu S,Wu T,et al.Detection of morphology defects in pipeline based on 3D active stereo omnidirectional vision sensor[J].Iet Image Processing,2018,12(4):588-595.
[18]Tang Yi-ping,Wu Ting,Yuan Gong-ping,et al.A 3D omni-directional vision sensor for detection of morphology defects in pipelines[J].Journal of Instrument and Meter,2017,38(3):726-733.
[1]郭伟青,汤一平,鲁少辉,等.基于镜面成像技术的三维立体视觉测量与重构综述[J].计算机科学,2016,43(9):1-10.
[2]汤一平.物联网感知技术与应用[M].北京:电子工业出版社,2015.
[12]束搏,邱显杰,王兆其.基于图像的几何建模技术综述[J].计算机研究与发展,2010,47(3):549-560.
[15]汤一平,杨瑞达,周伟敏,等.基于单目多视角机器视觉的真人鞋型复制装置和鞋楦制造方法:CN104573180A[P].2015-04-29.
[16]郭伟青,汤一平,夏少杰,等.基于镜面折反射全景成像的三维重建方法研究[J].高技术通讯,2015,25(z2):878-885.
[18]汤一平,吴挺,袁公萍,等.适用于管道内形貌检测的3D全景视觉传感器[J].仪器仪表学报,2017,38(3):726-733.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |