Relaxation Strategies for Feature Matching in Synthetic Aperture Radar (SAR) Images of Rural Scenes

Adrian N. Evans, Richard Lee Wilson, Edwin R. Hancock · 1995

A scheme is proposed that extracts linear features from rural SAR images, corresponding to hedgerow structures, and using a Delaunay graph to establish contextual neighbourhoods, produces successful matches with corresponding map data. This is achieved by the use of a hierarchical probabilistic relaxation scheme, that draws on a Bayesian model of the scene corruption produced by noise and segmentation error.

Read the paper · More papers on PaperTik