Community Research and Development Information Service - CORDIS

Abstract

Two examples are given of how the spatial analysis of remote sensing data can help to improve the classification of remotely sensed pictures, and to extract zones with similar parcel boundary characteristics. The first example is a digitised aerial photograph from a part of the city of Bangkok, Thailand. A classification on structural parameters, derived from an edge segmentation, supported in the last stage by spectral information, leads to better results than a per pixel classification. The second example is a Thematic Mapper image from an area nearby the city of Freiburg, F.R. Germany. Channel 5 is segmented and a region adjacency list is created. New images are created by assigning each region with a scaled structural property value. In some cases the values of properties of regions are integrated over the values of their neighbors to eliminate isolated peaks or drop outs. The combination of new images enhances structural information useful for geomorphologists, geologists and land use management. It seems that Thematic Mapper images have valuable intrinsic properties which can be used for extracting spatial homogeneous zones.

Additional information

Authors: STAKENBORG J H T JRC ISPRA ESTAB. (ITALY), JRC ISPRA ESTAB. (ITALY)
Bibliographic Reference: 1987 INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM, ANN ARBOR, MICHIGAN (USA), MAY 18-21, 1987 WRITE TO CEC LUXEMBOURG, DG XIII/A2, POB 1907 MENTIONING PAPER E 33432 ORA
Availability: Can be ordered online
Record Number: 1989126021201 / Last updated on: 1989-01-01
Category: PUBLICATION
Available languages: en