结合图像特征的多视拼接数据的消冗处理
|
摘要
在不丢失非重叠区域数据的情况下,对多视拼接重合区域的三维点云数据进行消冗处理是三维拼接中的一个难题。针对这一难题,提出了一种结合二维图像信息查找和消除冗余数据的新方法。算法首先查找位于拼接重叠区域的三维点云数据,结合三维点的K近邻约束和对应像素特征描述的相似度,对重叠区域的点云数据进行冗余查找和消除。实验表明,该方法能够准确判断并消除冗余点,没有造成更改或丢失非重叠区域三维数据点的不良效果,消冗速度也有所提高。
It is a difficult problem to make a redundancy processing for 3D point of overlapping areas after multi-view registration without any loss of data of non-overlapping areas.For this problem,this paper put forward a new algorithm,which found and eliminated the redundant data of overlapping areas combined with 2D image information.Firstly,it found the 3D point located in overlapping areas.Then it found and eliminated the redundant point of overlapping areas based on the constraints of K-nearest neighbors and the similarity of descriptors of corresponding pixels.Experimental results show that,the proposed method can estimate and eliminate the redundant point accurately,does not bring about any changing or lossing to the 3D point of non-overlapping areas,the speed is also increased.
It is a difficult problem to make a redundancy processing for 3D point of overlapping areas after multi-view registration without any loss of data of non-overlapping areas.For this problem,this paper put forward a new algorithm,which found and eliminated the redundant data of overlapping areas combined with 2D image information.Firstly,it found the 3D point located in overlapping areas.Then it found and eliminated the redundant point of overlapping areas based on the constraints of K-nearest neighbors and the similarity of descriptors of corresponding pixels.Experimental results show that,the proposed method can estimate and eliminate the redundant point accurately,does not bring about any changing or lossing to the 3D point of non-overlapping areas,the speed is also increased.
引文
[1]CHU Jun,LIU Xiao-ping,JIAO Chun-lin,et al.Multi-view reconstruc-tion of annular outdoor scenes from binocular video using globalrelaxation iteration[J].International Journal of Robotics and Au-tomation,2011,26(3):272-281.
[2]ZHAO Hui-jing,LIU Yi-ming,ZHU Xiao-long,et al.Scene under-standing in a large dynamic environment through a laser-based sensing[C]//Proc of IEEE International Conference on Robotics and Auto-mation.Piscataway:IEEE Press,2010:127-133.
[3]KOWDLE A,CHANG Y J,GALLAGHER A,et al.Active learning forpiecewise planar 3D reconstruction[C]//Proc of IEEE Conference onComputer Vision and Pattern Recognition.Washington DC:IEEEComputer Society,2011:929-936.
[4]刘晓茜,曹卫群,张璐.3维复杂树木场景建模及动画制作[J].中国图象图形学报,2010,15(1):136-141.
[5]BESL P J,MCKAY N D.A method for registration of 3D shapes[J].IEEE Trans on Pattern Analysis and Machine Intelligence,1992,14(2):239-256.
[6]方昭江,丁明跃.自由形态物体多视点深度数据配准方法研究[J].中国图象图形学报,2000,5(7):616-621.
[7]朱延娟,周来水,张丽艳.散乱点云数据配准算法[J].计算机辅助设计与图形学学报,2006,18(4):475-481.
[8]储珺,聂春梅,王璐,等.基于SIFT特征的多视点云数据配准和拼接算法[J].半导体光电,2011,32(3):442-447.
[9]LIU Zheng-jun,ZUO Zhi-quan,SONG Yuan-long,et al.Automaticpoint cloud data co-registration and overlap removal methods for highprecision terrain modeling and visualization[EB/OL].(2011-09-27).http://heritage.whu.edu.cn/lploadfile/dh 2011/Liu%20Zheng-jun.pdf.
[10]盛业华,张凯,张卡.多站拼接后三维激光扫描点云的消冗处理[J].测绘通报,2010(3):28-37.
[11]周儒荣,张丽艳,苏旭,等.海量散乱点的曲面重建算法研究[J].软件学报,2001,12(1):249-255.
[12]TOLA E,LEPETIT V,FUA P,et al.DAISY:an efficient dense de-scriptor applied to wide-baseline stereo[J].IEEE Trans on PatternAnalysis and Machine Intelligence,2010,32(5):815-830.
[13]MIKOLAJCZYK K,SCHMID C.A performance evaluation of local de-scriptors[J].IEEE Trans on Pattern Analysis and Machine Intel-ligence,2005,27(10):1615-1630.
[14]LOWE D G.Object recognition from local scale-invariant features[C]//Proc of the 7th IEEE International Conference on Computer Vi-sion.Piscataway:IEEE Press,1999:1150-1157.
[15]雷兰一菲,郎海涛.几种典型局部图像特征的比较[J].计算机应用,2010,30(S2):50-53.
[2]ZHAO Hui-jing,LIU Yi-ming,ZHU Xiao-long,et al.Scene under-standing in a large dynamic environment through a laser-based sensing[C]//Proc of IEEE International Conference on Robotics and Auto-mation.Piscataway:IEEE Press,2010:127-133.
[3]KOWDLE A,CHANG Y J,GALLAGHER A,et al.Active learning forpiecewise planar 3D reconstruction[C]//Proc of IEEE Conference onComputer Vision and Pattern Recognition.Washington DC:IEEEComputer Society,2011:929-936.
[4]刘晓茜,曹卫群,张璐.3维复杂树木场景建模及动画制作[J].中国图象图形学报,2010,15(1):136-141.
[5]BESL P J,MCKAY N D.A method for registration of 3D shapes[J].IEEE Trans on Pattern Analysis and Machine Intelligence,1992,14(2):239-256.
[6]方昭江,丁明跃.自由形态物体多视点深度数据配准方法研究[J].中国图象图形学报,2000,5(7):616-621.
[7]朱延娟,周来水,张丽艳.散乱点云数据配准算法[J].计算机辅助设计与图形学学报,2006,18(4):475-481.
[8]储珺,聂春梅,王璐,等.基于SIFT特征的多视点云数据配准和拼接算法[J].半导体光电,2011,32(3):442-447.
[9]LIU Zheng-jun,ZUO Zhi-quan,SONG Yuan-long,et al.Automaticpoint cloud data co-registration and overlap removal methods for highprecision terrain modeling and visualization[EB/OL].(2011-09-27).http://heritage.whu.edu.cn/lploadfile/dh 2011/Liu%20Zheng-jun.pdf.
[10]盛业华,张凯,张卡.多站拼接后三维激光扫描点云的消冗处理[J].测绘通报,2010(3):28-37.
[11]周儒荣,张丽艳,苏旭,等.海量散乱点的曲面重建算法研究[J].软件学报,2001,12(1):249-255.
[12]TOLA E,LEPETIT V,FUA P,et al.DAISY:an efficient dense de-scriptor applied to wide-baseline stereo[J].IEEE Trans on PatternAnalysis and Machine Intelligence,2010,32(5):815-830.
[13]MIKOLAJCZYK K,SCHMID C.A performance evaluation of local de-scriptors[J].IEEE Trans on Pattern Analysis and Machine Intel-ligence,2005,27(10):1615-1630.
[14]LOWE D G.Object recognition from local scale-invariant features[C]//Proc of the 7th IEEE International Conference on Computer Vi-sion.Piscataway:IEEE Press,1999:1150-1157.
[15]雷兰一菲,郎海涛.几种典型局部图像特征的比较[J].计算机应用,2010,30(S2):50-53.