Ponomarev S.V. Study of edge detection techniques in 3D image matching problem. Scientific and Technical Journal of Information Technologies, Mechanics and Optics
, 2017, vol. 17, no. 4, pp. 694–701 (in Russian). doi: 10.17586/2226-1494-2017-17-4-694-701
The paper deals withthe problem of image matching in three-dimensional space using contour description. An object-independent hierarchical structural juxtaposition algorithm is proposed. It is based on 2D structural matching algorithm using an alphabet of simple object-independent contour structural elements. This algorithm proved to be sufficiently robust and reliable for matching successfully the pictures of natural landscapes taken in differing seasons from differing aspect angles by differing sensors (the visible optical, IR, and SAR pictures, as well as the depth maps and geographical vector-type maps), but was unable to compare images of three-dimensional scenes, where it is required to apply different models of geometric transformations to different parts of the image. The three-dimensional version of the algorithm gives the possibility to overcome this limitation, but it is less robust with relation to changes in aspect angles. One of the key stages of the presented algorithm is the building of 3D scene contour descriptions. In order to increase the robustness of the algorithm a study of 3D edge detection techniques is carried out. The quantitative estimation of accuracy and speed of the presented techniques is given. The modification of the 3D structural matching algorithm based on the results of the study is proposed. The developed technique can be used for automatic navigation of extremely low flying unmanned vehicles or autonomous terrestrial robots in view of conditions with high degree a priori scene uncertainty.
structural matching, image alignment, edge detection, three-dimensional space Acknowledgements.
The work was supported by the RF Ministry of Education and Science, and partially by the Government of the Russian Federation (grant 074-U01). References
1. Booij O., Terwijn B., Zivkovic Z., Krose B. Navigation using an appearance based topological map. Proc. IEEE Int. Conf. on Robotics and Automation
, 2007, pp. 3927–3932. doi: 10.1109/ROBOT.2007.364081
2. Makadia A., Visontai M., Daniilidis K. Harmonic silhouette matching for 3D models. Proc. Int. Conf. on 3D TV
. Kos, Greece, 2007. doi: 10.1109/3DTV.2007.4379399
3. Papazov C., Burschka D. An efficient RANSAC for 3D object recognition in noisy and occluded scenes. Lecture Notes in Computer Science
, 2011, vol. 6492, pp. 135–148. doi: 10.1007/978-3-642-19315-6_11
4. Huang J., You S. Point cloud matching based on 3D self-similarity. Proc. IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops
, 2012, pp. 41–48. doi: 10.1109/CVPRW.2012.6238913
5. Flitton G., Breckon T.P., Megherbi N. Object recognition using 3D SIFT in complex CT volumes. Proc. British Machine Vision Conference
. Aberystwyth, UK,2010. doi: 10.5244/C.24.11
6. Lutsiv V.R. Object-independent approach to the structural analysis of images. Journal of Optical Technology, 2008, vol. 75, no. 11, pp. 708–714.
7. Andreev V.S., Iljashenko A.S., Kadykov A.B., Lapina N.N., Lutsiv V.R., Malyshev I.A., Novikova T.A., Potapov A.S., Gubkin A.F. Algorithms for automatically processing and analyzing aerospace pictures. Journal of Optical Technology, 2007, vol. 74, no. 5, pp. 307–322.
8. Ponomarev S.V., Lutsiv V.R., Malyshev I.A. Automatic structural matching of 3D image data. Proceedings of SPIE
, 2015, vol. 9649, art. 96490M. doi: 10.1117/12.2194312
9. Fleischman S., Cohenor D., Silva T. Robust moving least-squares fitting with sharp features. ACM Transactions on Graphics
, 2005, vol. 24, no. 3, pp. 544–552. doi: 10.1145/1073204.1073227
10. Lin Y., Wang C., Cheng J., Chen B., Jia C., Chen Z., Li J. Line segment extraction for large scale unorganized point clouds. ISPRS Journal of Photogrammetry and Remote Sensing
, 2015, vol. 102, pp. 172–183. doi: 10.1016/j.isprsjprs.2014.12.027
11. Hackel T., Wegner J.D., Schindler K. Contour detection in unstructured 3D point clouds. Proc. IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, USA, 2016, pp. 1610–1618.
12. Daniels J., Ochotta T., Ha L.K., Silva S.T. Spline-based feature curves from point-sampled geometry. Visual Computer
, 2008, vol. 24, no. 6, pp. 449–462. doi: 10.1007/s00371-008-0223-2
13. Hofer M., Maurer M., Bischof H. Line3D: efficient 3D scene abstraction for the built environment. Lecture Notes in Computer Science
, 2015, vol. 9358, pp. 237–248. doi: 10.1007/978-3-319-24947-6_19
14. Choi C., Trevor A.J. Christensen H.I. RGB-D edge detection and edge-based registration. Proc. 26th IEEE/RSJ Int. Conf. on Intelligent Robots and Systems
. Tokyo, Japan, 2013, pp. 1568–1575. doi: 10.1109/IROS.2013.6696558
15. Weber C., Hahmann S., Hagen H. Methods for feature detection in point clouds. Proc. Workshop on Visualization of Large and Unstructured Data Sets. Bodega Bay, USA, 2010, vol. 19, pp. 90–99.
16. Monga O., Deriche R., Malandain G., Cocquerez J.P. Recursive filtering and edge tracking: two primary tools for 3D edge detection. Image and Vision Computing
, 1991, vol. 9, no. 4,
pp. 203–214. doi:10.1016/0262-8856(91)90025-K
17. Hemmat J., Bondarev E., De With P.H.N. Real-time planar segmentation of depth images: from three-dimensional edges to segmented planes. Journal of Electronic Imaging
, 2015, vol. 24, no. 5, art. 051008. doi: 10.1117/1.JEI.24.5.051008
18. Zhou Q.-Y., Koltun V. Dense scene reconstruction with points of interest. ACM Transactions on Graphics
, 2014, vol. 33, no. 4.doi: 10.1145/2461912.2461919