Year: 2013
Type: conference-paper
Abstract: Pose uncertainty estimation of calibrated cameras is a common task in the field of computer vision and uses\nlocation uncertainties of image features. For spherical cameras, those uncertainties cannot be optimally described\nusing conventional latitude-longitude representation. Increasing distortions close to the poles of the spherical\ncoordinate system prevent a suitable description through Gaussians.\nTo overcome this limitation, we present a consistent location uncertainty representation for spherical image features:\nOur approach is based on normal vectors in Cartesian space and applicable to any kind of camera with\nconvex projection surfaces, such as catadioptric and spherical systems. We compare its performance against\nlatitude-longitude representation by estimating camera pose uncertainties through first order error propagation\nin a weighted least squares pose estimation scenario. Our experiments on synthetic and real data show that the\nproposed approach delivers consistent results outperforming conventional latitude-longitude representation.
Institution: University of Kaiserslautern
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2405653789