基于特征点和最优路径的无人机影像拼接方法
|
摘要
利用SIFT算法对无人机影像进行匹配,采用距离阈值的方法剔除匹配后距离较近的特征点。为了降低误差累积对影像拼接效果的影响,提出了最佳基准影像的选择方法,同时将Dijkstra算法引入到拼接路径的搜索中,利用影像间的投影转换误差构建搜索权阵。结果表明:最优路径的拼接方法能有效避免投影误差较大的影像对后续拼接影像的影响,减少了影像投影转换次数,改善了影像的拼接效果。
The SIFT algorithm was used in Unmanned Aerial Vehicle images matching,the distance threshold filter was used to remove too close feature points.In order to reduce the error accumulation which can effect on the results of image mosaic,and shorten the path splicing,an optimal base image selection method was introduced,meanwhile,the Dijkstra algorithm was introduced into the search of stitching path,and the project conversion errors between the two images were used to build the search array.The experimental results indicated that this optimal path method can avoid the influences of the image which has large projection error mosaic with the follow-up images effectively,and reduce the number of image projection transformation,which can improve the mosaic effect.
The SIFT algorithm was used in Unmanned Aerial Vehicle images matching,the distance threshold filter was used to remove too close feature points.In order to reduce the error accumulation which can effect on the results of image mosaic,and shorten the path splicing,an optimal base image selection method was introduced,meanwhile,the Dijkstra algorithm was introduced into the search of stitching path,and the project conversion errors between the two images were used to build the search array.The experimental results indicated that this optimal path method can avoid the influences of the image which has large projection error mosaic with the follow-up images effectively,and reduce the number of image projection transformation,which can improve the mosaic effect.
引文
[1]Lu Heng,Li Yongshu,Li Hechao,et al.Digital Processing ofUnmanned Aerial Vehicle Image and Its Application in Re-construction of Wenchuan Earthquake-hit Areas[J].Journalof Southwest Jiaotong University,2010,45(4):533-538.[鲁恒,李永树,李何超,等.无人机影像数字处理及在地震灾区重建中的应用[J].西南交通大学学报:自然科学版,2010,45(4):533-538.]
[2]Zhong Li,Hu Xiaofeng.Stitching Algorithm for OverlappingImags[J].Journal of Image and Graphics,1998,5(5):367-370.[钟力,胡晓峰.重叠图像拼接算法[J].中国图象图形学报,1998,5(5):367-370.]
[3]Zhao Xiangyang,Du Limin.An Automatic and Robust ImageMosaic Algorithm[J].Journal of Image and Graphics,2004,9(4):417-422.[赵向阳,杜利民.一种全自动稳健的图像拼接融合算法[J].中国图象图形学报,2004,9(4):417-422.]
[4]Reddy B S,Chatterji B N.An FFT-based Technique forTranslation,Rotation,and Scale-Invariant Image Registration[J].IEEE Transaction on Image Processing,1996,5(8):1266-1271.
[5]Szeliski R.Video Mosaics for Virtual Environments[J].IEEEComputer Graphics and Applications,1996,16(2):22-30.
[6]Peleg S,Rousso B,Rav-Achaetc A.Mosaic on Adaptive Mani-folds[J].IEEE Transaction on Pattern Analysis and MachineIntelligence,2000,22(10):1144-1154.
[7]Brown M,Lowe D G.Recognizing Panoramas[C]//Proceed-ings of the IEEE Ninth International Conference on ComputerVision,Washington,DC,USA,2003,2:1218-1225.
[8]Lowe D G.Object Recognition from Local Scale-InvariantFeatures[C]//International Conference on Computer Vision,Corfu,Greece,1999,2:1150-1157.
[9]Lowe D G.Distinctive Image Features from Scale-InvariantKeypoints[J].International Journal of Computer Vision,2004,60(2):91-110.
[10]Mikolajczyk K,Leibe B,Schiele B.Local Features for ObjectClass Recognition[C]//Proceedings of the IEEE Ninth Inter-national Conference on Computer Vision,Beijing,China,2005,2:1792-1799.
[11]Yang Changbao,Wang Kaiyi,Ma Shengzhong.Implementa-tion of an Optimized Algorithms in the Shortest Path Analy-sis[J].Journal of Jilin University(Information Science Edi-tion),2002,20(2):70-74.[杨长保,王开义,马生忠.一种最短路径分析优化算法的实现[J].吉林大学学报(信息科学版),2002,20(2):70-74.]
[2]Zhong Li,Hu Xiaofeng.Stitching Algorithm for OverlappingImags[J].Journal of Image and Graphics,1998,5(5):367-370.[钟力,胡晓峰.重叠图像拼接算法[J].中国图象图形学报,1998,5(5):367-370.]
[3]Zhao Xiangyang,Du Limin.An Automatic and Robust ImageMosaic Algorithm[J].Journal of Image and Graphics,2004,9(4):417-422.[赵向阳,杜利民.一种全自动稳健的图像拼接融合算法[J].中国图象图形学报,2004,9(4):417-422.]
[4]Reddy B S,Chatterji B N.An FFT-based Technique forTranslation,Rotation,and Scale-Invariant Image Registration[J].IEEE Transaction on Image Processing,1996,5(8):1266-1271.
[5]Szeliski R.Video Mosaics for Virtual Environments[J].IEEEComputer Graphics and Applications,1996,16(2):22-30.
[6]Peleg S,Rousso B,Rav-Achaetc A.Mosaic on Adaptive Mani-folds[J].IEEE Transaction on Pattern Analysis and MachineIntelligence,2000,22(10):1144-1154.
[7]Brown M,Lowe D G.Recognizing Panoramas[C]//Proceed-ings of the IEEE Ninth International Conference on ComputerVision,Washington,DC,USA,2003,2:1218-1225.
[8]Lowe D G.Object Recognition from Local Scale-InvariantFeatures[C]//International Conference on Computer Vision,Corfu,Greece,1999,2:1150-1157.
[9]Lowe D G.Distinctive Image Features from Scale-InvariantKeypoints[J].International Journal of Computer Vision,2004,60(2):91-110.
[10]Mikolajczyk K,Leibe B,Schiele B.Local Features for ObjectClass Recognition[C]//Proceedings of the IEEE Ninth Inter-national Conference on Computer Vision,Beijing,China,2005,2:1792-1799.
[11]Yang Changbao,Wang Kaiyi,Ma Shengzhong.Implementa-tion of an Optimized Algorithms in the Shortest Path Analy-sis[J].Journal of Jilin University(Information Science Edi-tion),2002,20(2):70-74.[杨长保,王开义,马生忠.一种最短路径分析优化算法的实现[J].吉林大学学报(信息科学版),2002,20(2):70-74.]
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心