ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., I-3, 57-61, 2012
http://www.isprs-ann-photogramm-remote-sens-spatial-inf-sci.net/I-3/57/2012/
doi:10.5194/isprsannals-I-3-57-2012
© Author(s) 2012. This work is distributed
under the Creative Commons Attribution 3.0 License.
 
19 Jul 2012
RELATIVE POSE ESTIMATION FROM AIRBORNE IMAGE SEQUENCES
T. Reize, R. Müller, and F. Kurz The Remote Sensing Technology Institute, German Aerospace Center (DLR), Oberpfaffenhofen, Germany
Keywords: Photogrammetry, Image Orientation Estimation Method, Matching, Algorithms, IMU Abstract. We present a new relative pose estimation method for applications based on airborne image sequences. The performance of the method is tested using simulated test data, with correct and erroneous original conditions, as well as using real data. The calculated results obtained from real images are compared to the on-board measured angles. The results show that the proposed method is very precise and fast. Most matching algorithms are very computation time expensive mainly because they rely on RANSAC methods that need a lot of matching points. Due to the circumstance that only two corresponding points are necessary to solve the equation system, our technique doesn't need much computation time. Outliers are detected by a special back-matching technique. A method based on Polynomial Homotopy Continuation (PHC) is used to solve the complex polynomial equation system. The proposed pose solver method runs without SVD calculations, expensive minimisation or optimisation. Start parameters are not necessary. Furthermore, no a priori knowledge is required, besides focal length in pixel units and overlapping consecutive images. Outcomes are three relative orientation angles and a scaling parameter between two subsequent images, as well as displacement vectors in image pixel coordinate units. In addition, the PHC pose estimation method can balance small pixel errors. All these properties indicate the high applicability of the proposed method.
Conference paper (PDF, 756 KB)


Citation: Reize, T., Müller, R., and Kurz, F.: RELATIVE POSE ESTIMATION FROM AIRBORNE IMAGE SEQUENCES, ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., I-3, 57-61, doi:10.5194/isprsannals-I-3-57-2012, 2012.

BibTeX EndNote Reference Manager XML