Panoramic volumes for robot localization

We propose a method for visual robot localization using a panoramic image volume as the representation from which we can generate views from virtual viewpoints and match them to the current view. We use a geometric image-based rendering formalism in combination with a subspace representation of images, which allows us to synthesize views at arbitrary virtual viewpoints from a compact low-dimensional representation.

Authors

• Matej Artac
• Matjaz Jogan
• Ales Leonardis
• Hynek Bakstein

Keywords

• Robot localization
• Rendering (computer graphics)
• Solid modeling
• Computer vision
• Robot vision systems
• Cameras
• Mobile robots
• Image storage
• Information science
• Cybernetics
• Image Volumes
• Image Representation
• Low-dimensional Representation
• Compact Representation
• Virtually
• Similarity Measure
• Column Vector
• Image Reconstruction
• Focal Length
• Reference Image
• Local Estimates
• Depth Map
• Mobile Robot
• Dimensional Representation
• Image Synthesis
• Global Representation
• Position Of The Robot
• View Of Activity
• Artificial Sequences
• Straight Trajectory
• Virtual Camera
• Images In Column
• Images In Row
• Eigenspace
• View Synthesis
• Global Minimum
• Changes In Appearance
• visual localization
• panoramic images
• subspace representation