Model Based Pose Estimation Using SURF
Estimation of a camera pose (position and orientation) from an image, given a 3d model of the world, is a topic of great interest in many current fields of research. When aiming for a model based pose estimation approach, several questions arise: What is the model? How do we acquire a model? How is the image linked to the model? How is a pose computed and verified using the latter information? In this paper we present a new approach towards model based pose estimation based solely on SURF features. We give a formal definition of our model, show how to build such a model from image data automatically, how to integrate two partial models, and how pose estimation for new images works.
Unable to display preview. Download preview PDF.
- 1.Bay, H., Ess, A., Tuytelaars, T., van Gool, L.: Speeded-up robust features (surf). Journal of Computer Vision 110, 346–359 (2008)Google Scholar
- 3.Schindler, G., Brown, M., Szeliski, R.: City-scale location recognition. In: IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp. 1–7 (2007)Google Scholar
- 4.Wu, C., Fraundorfer, F., Frahm, J.M., Pollefeys, M.: 3d model search and pose estimation from single images using vip features. In: Computer Vision and Pattern Recognition Workshop, pp. 1–8 (2008)Google Scholar
- 6.Irschara, A., Zach, C., Frahm, J., Bischof, H.: From structure-from-motion point clouds to fast location recognition, pp. 2599–2606 (2009)Google Scholar
- 10.Nistér, D.: An efficient solution to the five-point relative pose problem. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 195–202 (2003)Google Scholar
- 12.Lourakis, M.: levmar: Levenberg-marquardt nonlinear least squares algorithms in c/c++ (2004) (accessed on January 31, 2005)Google Scholar
- 13.Muja, M.: Flann, fast library for approximate nearest neighbors (2009), http://mloss.org/software/view/143/