Perspectively Invariant Normal Features

We extend the successful 2D robust feature concept into the third dimension in that we produce a descriptor for a reconstructed 3D surface region. The descriptor is perspectively invariant if the region can locally be approximated well by a plane. We exploit depth and texture information, which is nowadays available in real-time from video of moving cameras, from stereo systems or PMD cameras (photonic mixer devices). By computing a normal view onto the surface we still keep the descriptiveness of similarity invariant features like SIFT while achieving in- variance against perspective distortions, while descriptiveness typically suffers when using affine invariant features. Our approach can be exploited for structure-from-motion, for stereo or PMD cameras, alignment of large scale reconstructions or improved video registration.

[1]  David G. Stork,et al.  Pattern classification, 2nd Edition , 2000 .

[2]  Cordelia Schmid,et al.  A Comparison of Affine Region Detectors , 2005, International Journal of Computer Vision.

[3]  David G. Stork,et al.  Pattern Classification , 1973 .

[4]  Horst Bischof,et al.  A novel performance evaluation method of local detectors on non-planar scenes , 2005, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) - Workshops.

[5]  Reinhard Koch,et al.  Visual Modeling with a Hand-Held Camera , 2004, International Journal of Computer Vision.

[6]  Reinhard Koch,et al.  A Comparison of PMD-Cameras and Stereo-Vision for the Task of Surface Reconstruction using Patchlets , 2007, 2007 IEEE Conference on Computer Vision and Pattern Recognition.

[7]  Margaret M. Fleck.  Perspective Projection: The Wrong Imaging Model , 1995 .

[8]  Shigeo Abe DrEng.  Pattern Classification , 2001, Springer London.

[9]  Cordelia Schmid,et al.  A Performance Evaluation of Local Descriptors , 2005, IEEE Trans. Pattern Anal. Mach. Intell..

[10]  Roland Siegwart,et al.  A Flexible Technique for Accurate Omnidirectional Camera Calibration and Structure from Motion , 2006, Fourth IEEE International Conference on Computer Vision Systems (ICVS'06).

[11]  G LoweDavid,et al.  Distinctive Image Features from Scale-Invariant Keypoints , 2004 .

[12]  X. Shi,et al.  Invariant operators, small samples, and the bias-variance dilemma , 2004, CVPR 2004.

[13]  Vincent Lepetit,et al.  Randomized trees for real-time keypoint recognition , 2005, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05).

[14]  Horst Bischof,et al.  Efficient Maximally Stable Extremal Region (MSER) Tracking , 2006, 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'06).

[15]  Reinhard Koch,et al.  Robust Feature Representation for Efficient Camera Registration , 2006, DAGM-Symposium.

[16]  Zhaohui Wu,et al.  Sphere-Spin-Image: A Viewpoint-Invariant Surface Representation for 3D Face Recognition , 2004, International Conference on Computational Science.

[17]  Jean-Yves Bouguet,et al.  Camera calibration toolbox for matlab , 2001 .

[18]  Aly A. Farag,et al.  Surfacing Signatures: An Orientation Independent Free-Form Surface Representation Scheme for the Purpose of Objects Registration and Matching , 2002, IEEE Trans. Pattern Anal. Mach. Intell..

[19]  Minglun Gong,et al.  Fast Unambiguous Stereo Matching Using Reliability-Based Dynamic Programming , 2005, IEEE Trans. Pattern Anal. Mach. Intell..

[20]  Jiri Matas,et al.  Robust wide-baseline stereo from maximally stable extremal regions , 2004, Image Vis. Comput..