Visual Place Representation and Recognition from Depth Images
This addresses place recognition for robotics or AR applications, but appears incremental as it builds on existing depth-based methods.
The paper tackles place recognition from depth images by computing 3D models and deriving 2D maps to create scene descriptors, showing efficiency and robustness to appearance changes and light variations.
This work proposes a new method for place recognition based on the scene architecture. From depth video, we compute the 3D model and we derive and describe geometrically the 2D map from which the scene descriptor is deduced to constitute the core of the proposed algorithm. The obtained results show the efficiency and the robustness of the propounded descriptor to scene appearance changes and light variations.