The new generation of artificial satellites is providing a huge amount of Earthudobservation images whose exploitation can report invaluable benefits, both economical andudenvironmental. However, only a small fraction of this data volume has been analyzed, mainlyuddue to the large human resources needed for that task. In this sense, the development ofudunsupervised methodologies for the analysis of these images is a priority. In this work, audnew unsupervised segmentation algorithm for satellite images is proposed. This algorithm isudbased on the rough-set theory, and it is inspired by a previous segmentation algorithm definedudin the RGB color domain. The main contributions of the new algorithm are: (i) extendingudthe original algorithm to four spectral bands; (ii) the concept of the superpixel is used inudorder to define the neighborhood similarity of a pixel adapted to the local characteristics ofudeach image; (iii) and two new region merged strategies are proposed and evaluated in orderudto establish the final number of regions in the segmented image. The experimental resultsudshow that the proposed approach improves the results provided by the original method whenudboth are applied to satellite images with different spectral and spatial resolutions.
展开▼
