Remote mining: from clustering to DTM

Description

LIDAR data acquisition is becoming an indispensable task for terrain characterization in large surfaces. In Mediterranean woods this job results hard due to the great variety of heights and forms, as well as sparse vegetation that they present. A new data mining-based approach is proposed with the aim of classifying LIDAR data clouds as a first step in DTM generation. The developed methodology consists in a multi-step iterative process that splits the data into different classes (ground and low/med/high vegetation) by means of a clustering algorithm. This method has been tested on three different areas of the southern Spain with successful results, verging on 80% hits

Abstract

Ministerio de Ciencia y Tecnología TIN2007-68084

Additional details