Image Filtering Based on Locally Estimated Geodesic Functions

This paper addresses the problem of edge-preserving smoothing of natural images. A novel adaptive approach as a preprocessing stage in feature extraction and/or image segmentation. It performs a weighted convolution by combining both spatial and tonal information in a single similarity measure based on the local calculation of geodesic time functions. Two different strategies are derived for smoothing heterogeneous areas while preserving relevant structures.JRC.DDG.H.6-Spatial data infrastructure