A Hybrid Method for Segmentation of Multispectral Remote Sensing Images

D. Sastre, N. Snchez, J. Portillo, G. Cisneros, and J.M. Menndez (Spain)


Segmentation, texture, edge detection, remote sensing imagery.


Remote sensing images often present a hierarchical structure, composed of a variety of nested regions each of which represents a different area: water basins, agricultural fields, urban regions… In several applications such as cadastre information retrieval it is important to identify and distinguish these regions in an easy and automatic way. Region-based segmentation approaches allow to achieve quite good results. However, in the field of our application, a higher level of performance is required. Consequently, a novel hybrid method which combines region segmentation based on textures and edge detection is proposed, obtaining much better results than those achieved by applying each process individually.

Important Links:

Go Back