®®®® SIIA Público

Título del libro: 2006 Ieee International Geoscience And Remote Sensing Symposium, Vols 1-8
Título del capítulo: Stochastic scheme to street detection in remote sensing images

Autores UNAM:
MIGUEL MOCTEZUMA FLORES;
Autores externos:

Idioma:

Año de publicación:
2006
Palabras clave:

ROADS


Resumen:

We propose a four-step segmentation scheme to street detection. To avoid long computing time, in the first step the algorithm is based in maximum-likelihood framework. It provides a fast map of the streets, but is a rough approximation. In the second step we define a stochastic segmentation stage by means of MRF theory. The goal is to classify the rejected class regions. Our algorithm is intended to provide segmented images preserving slight details. In order to trace the streets and get the center and connected nodes we used a morphological operator. Final step defines a parametric algorithm to detect straight lines. The scheme is applied in panchromatic aerial images.


Entidades citadas de la UNAM: