基于联合SIFT和SURF特征的三维表面重建
|
摘要
一般提取二维图像特征点的方法用到SIFT特征提取,因为SIFT特征有几个特性:对噪声和光线容忍度高、区分性、多量性、可扩展性等,但对于边缘光滑目标的特征点提取能力较弱。SURF特征也是提取图像的尺度不变特征,SURF方法使用Hessian矩阵的行列式值作特征点检测,在对于光滑边缘的目标特征点检测效果要优于SIFT特征。采用同时提取图像中SIFT和SURF特征的方法用于关键点的确定,能够在SIFT特征稳定性好、尺度不变性基础上,提高边缘光滑目标的特征点检测能力。实验结果表明,使用SIFT特征和SURF特征联合的方法能够重建出更多的顶点数和面片数,包括利用SIFT特征提取后存在空缺的部分。重建出的三维表面有更完整更准确的顶点和三角形面片,能提高重建表面的完整度与真实性。
The general method of extracting 2 D image feature points uses SIFT feature extraction because SIFT features have several characteristics: high tolerance to noise and light, discriminating, multiplicity, scalability, etc., but the extraction ability for feature points of edge smooth targets is weak. The SURF feature is also a scale-invariant feature of the extracted image. SURF method uses the determinant value of the Hessian matrix for feature point detection, and the detection effect on the target feature point for the smooth edge is better than the SIFT feature. In this paper, the method of simultaneously extracting the SIFT and SURF features in the image is used to determine the key points. Based on the SIFT feature stability and scale invariance, the feature point detection ability of the edge smooth target can be improved. The experimental result expression, using jointly SIFT features and SURF features, can obtain more number of vertexes and patches after reconstruction of point clouds, including the vacancy part after the SIFT feature extraction. The increase of point cloud information provides more complete and accurate point cloud information for the 3 D surface reconstruction step, which can improve the integrity and authenticity of the reconstructed surface.
The general method of extracting 2 D image feature points uses SIFT feature extraction because SIFT features have several characteristics: high tolerance to noise and light, discriminating, multiplicity, scalability, etc., but the extraction ability for feature points of edge smooth targets is weak. The SURF feature is also a scale-invariant feature of the extracted image. SURF method uses the determinant value of the Hessian matrix for feature point detection, and the detection effect on the target feature point for the smooth edge is better than the SIFT feature. In this paper, the method of simultaneously extracting the SIFT and SURF features in the image is used to determine the key points. Based on the SIFT feature stability and scale invariance, the feature point detection ability of the edge smooth target can be improved. The experimental result expression, using jointly SIFT features and SURF features, can obtain more number of vertexes and patches after reconstruction of point clouds, including the vacancy part after the SIFT feature extraction. The increase of point cloud information provides more complete and accurate point cloud information for the 3 D surface reconstruction step, which can improve the integrity and authenticity of the reconstructed surface.
引文
[1] 吴彤,傅中力.三维重建技术及其军事应用.国防科技,2015,36(1):31-34.
[2] 曾宁,范应方,杨剑,等.数字虚拟技术在肝胆外科临床教学中的应用研究[J].中国继续医学教育,2018,10(33):16-19.
[3] 明国辉,委民正.SURF算法在无人机倾斜摄影测量三维建模中的应用[J].测绘工程,2017,26(9):41-45.
[4] 吴宁,陈佳舟,吴凯乐.面向数字化保护的自动文物三维重建方法研究[J].山西建筑,2018,44(6):257-258.
[5] 霍林生,张耀文,李宏男.图像三维重建法在震损建筑实体建模中的应用研究[J].世界地震工程,2017,33(2):113-118.
[6] SEITZ S M,CURLESS B,DIEBEL J,et al.A comparison and evaluation of multi-view stereo reconstruction algorithms[C].Ppwer Symposium (NAPS),2005:519-528.
[7] FUHRMANN S,LANGGUTH F,GOESELE M.MVE-A multiview reconstruction environment[C].GCH′14 Proceedings of the Eurogcaphics Workshop on Graphics and Cultural Heritage,2014:11-18.
[8] UMMENHOFER B,BROX T.Global,dense multiscale reconstruction for a billion points[C].Proceedings of the IEEE International Conference on Computer Vision.2015:1341-1349.
[9] LOWE D G.Distinctive image features from scale-invariant keypoints[J].International Journal of Computer Vision,2004,60(2):91-110.
[10] BAY H,ESS A,TUYTELAARS T,et al.Speeded-up robust features (SURF)[J].Computer Vision And Image Understanding,2008,110(3):346-359.
[11] MANICKAM A,DEVARASAN E,MANOGARAN G,et al.Score level based latent fingerprint enhancement and matching using SIFT feature[J].Multimedia Tools and Applications,2018,78(3):1-21.
[12] ELTNER A,KAISER A,CASTILLO C,et al.Image-based surface reconstruction in geomorphometry-merits,limits and developments[J].Earth Surface Dynamics,2016,4(2):359-389.
[13] JIN R,KIM J.Tracking feature extraction techniques with improved SIFT for video identification[J].Multimedia Tools and Applications,2017,76(4):5927-5936.
[14] 索春宝,杨东清,刘云鹏.多种角度比较SIFT、SURF、BRISK、ORB、FREAK算法[J].北京测绘,2014(4):23-26,22.
[15] 王凌云,尹海波,王琪.SURF和RANSAC在图像拼接中的应用[J].电子测量技术,2016,39(4):71-75.
[2] 曾宁,范应方,杨剑,等.数字虚拟技术在肝胆外科临床教学中的应用研究[J].中国继续医学教育,2018,10(33):16-19.
[3] 明国辉,委民正.SURF算法在无人机倾斜摄影测量三维建模中的应用[J].测绘工程,2017,26(9):41-45.
[4] 吴宁,陈佳舟,吴凯乐.面向数字化保护的自动文物三维重建方法研究[J].山西建筑,2018,44(6):257-258.
[5] 霍林生,张耀文,李宏男.图像三维重建法在震损建筑实体建模中的应用研究[J].世界地震工程,2017,33(2):113-118.
[6] SEITZ S M,CURLESS B,DIEBEL J,et al.A comparison and evaluation of multi-view stereo reconstruction algorithms[C].Ppwer Symposium (NAPS),2005:519-528.
[7] FUHRMANN S,LANGGUTH F,GOESELE M.MVE-A multiview reconstruction environment[C].GCH′14 Proceedings of the Eurogcaphics Workshop on Graphics and Cultural Heritage,2014:11-18.
[8] UMMENHOFER B,BROX T.Global,dense multiscale reconstruction for a billion points[C].Proceedings of the IEEE International Conference on Computer Vision.2015:1341-1349.
[9] LOWE D G.Distinctive image features from scale-invariant keypoints[J].International Journal of Computer Vision,2004,60(2):91-110.
[10] BAY H,ESS A,TUYTELAARS T,et al.Speeded-up robust features (SURF)[J].Computer Vision And Image Understanding,2008,110(3):346-359.
[11] MANICKAM A,DEVARASAN E,MANOGARAN G,et al.Score level based latent fingerprint enhancement and matching using SIFT feature[J].Multimedia Tools and Applications,2018,78(3):1-21.
[12] ELTNER A,KAISER A,CASTILLO C,et al.Image-based surface reconstruction in geomorphometry-merits,limits and developments[J].Earth Surface Dynamics,2016,4(2):359-389.
[13] JIN R,KIM J.Tracking feature extraction techniques with improved SIFT for video identification[J].Multimedia Tools and Applications,2017,76(4):5927-5936.
[14] 索春宝,杨东清,刘云鹏.多种角度比较SIFT、SURF、BRISK、ORB、FREAK算法[J].北京测绘,2014(4):23-26,22.
[15] 王凌云,尹海波,王琪.SURF和RANSAC在图像拼接中的应用[J].电子测量技术,2016,39(4):71-75.