In this work, we perform three-dimensional scene recovery from image data capturing railway transportation corridors. Typical three-dimensional scene recovery methods initialise recovered feature positions by searching for correspondences between image frames. We intend to take advantage of a relationship between image data and recovered scene data to reduce the search space traversed when performing such correspondence matching. We build multi-dimensional Gaussian models of recurrent visual features associated with distributions representing recovery results from our own dense planar recovery method. Results show that such a scheme decreases the number of checks made per feature to 6% of a comparable exhaustive method, whilst unaffecting accuracy. Further, the proposed method performs competitively when compared with other methods presented in literature. © 2011 Springer-Verlag.
CITATION STYLE
Warsop, T., & Singh, S. (2011). Unsupervised learning for improving efficiency of dense three-dimensional scene recovery in corridor mapping. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 6688 LNCS, pp. 393–402). https://doi.org/10.1007/978-3-642-21227-7_37
Mendeley helps you to discover research relevant for your work.