Last modified: 2016-05-10
Abstract
Unmanned Aerial Vehicle (UAV) attached with a non-metric camera is becoming a popular platform for acquisition of aerial images. It allows users to readily acquire geospatial data, with low cost. With rapid development of computer science and photogrammetry, generation of digital surface models, orthoimage, and 3D color point clouds become an automatic process. Some commercial software, such as Agisoft Photoscan or Pix4Dmapper, enable users to rapidly produce these UAV-derived products, without geomatics background needed. However, how to use the UAV-derived geospatial data to effectively investigate the change of the Earth surfaces over time is of importance. In other words, extracting useful information from low-cost geospatial data would further extend the advantages of employing UAV. This study makes good use of UAV-derived orthoimage and digital surface model (DSM) to identify where the change of the earth and the magnitude. A set of orthoimage and DSM are considered to be historical data, which are produced through standard photogrammetric procedure. The Ground Sampling Distance (GSD) for UAV-derived orthoimage and historical orthoimage is 13 cm and 25 cm; the grid size for UAV-derived DSM and historical DSM is 25 cm and 2 m.
An object-oriented analysis has been a popular method for digital image classification. The technique of image segmentation is employed to convert an image into multiple objects. In this study, we assign different weights to the orthoimages and the difference of the DSMs over time for the segmentation process. Such a strategy helps rapidly identify significant change of the earth. In addition, the magnitude of the change is estimated.