Please use this identifier to cite or link to this item:
|Title:||Updating an object-based pan-tropical forest cover change assessment by automatic change detection and classification|
|Authors:||BEUCHLE Rene'; RASI Rastislav; BODART Catherine; VOLLMAR MICHAEL; SELIGER Roman; ACHARD Frederic|
|Citation:||Proceedings of the 4th Conference on Geographic Object-Based Image Analysis p. 332-337|
|Publisher:||National Institute for Space Research (INPE)|
|Type:||Articles in periodicals and books|
|Abstract:||The TREES-3 project of the European Commission’s Joint Research Centre is producing estimates of tropical forest cover changes during the period 1990 to 2010. Three reference years are considered: 1990, 2000 and 2010. This paper presents the method developed for the automatic processing of year 2010 with the assessment of performance of this method. The processing of imagery of year 2010 includes automatic segmentation, change detection and object spectral classification. The validated maps of forest cover changes for the period 1990-2000 are used as thematic input layer into the segmentation and classification process of the year 2010 images. Object-based change detection (OBCD) technique is applied using Tasselled Cap (TCap) parameters and spectral Euclidian Distances (ED). Objects detected as changed are classified by change vector analysis. The segmentation approach was tested on a subsample of 568 sample units over Brazil. The segmentation results for year 2010 are consistent with segmentation of imagery for the period 1990-2000, the segmentation statistics (number of objects, average objects size, average number of objects per sample site) remain stable between the two approaches. A two-step method of (a) change detection and (b) classification of changed objects was developed on basis of thresholding TCap variance and Euclidian Distance. The approach was tested over 281 sample units in the Brazilian biome of the Amazon, for which validated land cover information for the year 2010 was already available. The resulting overall accuracy of classification for the 281 sample units was 92.2%.|
|JRC Institute:||Sustainable Resources|
Files in This Item:
There are no files associated with this item.
Items in repository are protected by copyright, with all rights reserved, unless otherwise indicated.