International Archives of Photogrammetry and Remote Sensing, Vol. 32, Part 7-4-3 W6, Valladolid, Spain, 3-4 June, 1999
184
SENSOR FUSED IMAGES FOR VISUAL INTERPRETATION OF FOREST STAND BORDERS
R. Fritz, I. Freeh, B. Koch, Chr. Ueffing
Dept, of Remote Sensing and LIS, University of Freiburg, Tennenbacherstr. 4, 79106 Freiburg, Germany,
feminfo@felis.uni-freiburg.de
KEYWORDS: IHS-Transformation, Adaptive Image Fusion, Forest Stand Border Delineation, Quality Analysis.
ABSTRACT
In the last years forest cartography is changing from an analog to a digital generation and presentation. For this purpose, forest
departments develop and implement Forest Geographic Information Systems. Till now, the actual stand borders are delineated based
on B/W orthophotos. This procedure requires at least four processing steps to integrate stand delineations assessed during forest
inventories into the final digital forest management map. Each step produces errors and makes the procedure of data integration
labour- and cost-intensive. With the availability of high resolution satellite data, the use of such data for forest map production
becomes feasible. Due to the information requirements, high spatial and spectral resolution is needed. The fusion of image data
provides an opportunity to meet these requirements. Different fusion techniques have been tested, such as IHS-, PC- and Brovey-
transformation to combine the spectral and spatial information of satellite data. All three sensor fusion methods show a visual
improvement of the images by the synergy effect combining high spectral and spatial satellite data, whereby the IHS transformation
showed best colour differentiation for visual interpretation.
Due to the good performance of the IHS transformed images, the applicability of sensor fusion techniques for forest inventory
mapping has been investigated with an IHS transformed product of Landsat TM and IRS-1C pan [IHS_TM]. This product was
compared to an IHS transformed SPOT XS and PAN [IHS_SP], IRS-1C pan [Pan] alone, a B/W orthophoto [Ortho] and a simulated
QuickBird image [Qsim] with lm resolution. The resulting image fusion products have been interpreted and digitized ‘on-screen’ by
two persons in order to delineate stand borders.
The comparative evaluation is performed in two steps: (1) measuring the visibility percentage of forest stand borders in reference to
the official forest inventory maps and (2) calculation of quality measure criteria. The total length of stand borders in the test area is
more than 40km. In every image product more than 70 % of stand borders have been detected, with IHS 72%, PAN 76% and
simulated QUICK 85 % correspondence.
Another tested approach was to combine Landsat TM and KVR-1000 (2m resolution) with different fusion techniques. In a first step,
an adaptive image fusion (AIF) was calculated. In a second step, the resulting image was combined with the high resolution data,
again with an IHS transformation. The output of this transformation shows very good results for visual interpretation and will be used
as photorealistic background for tourist information systems. A precondition of this method is a very accurate rectification of the
input data. Otherwise, it will provide poor results with a lot of artefacts.
1. INTRODUCTION
With further development of information technology the
requirements of forest departments for using geographic
information systems increased at the end of the 80s (Teuffel and
Krebs, 1996). The result of this development was the
implementation of Forest Geographic Information Systems. The
forest cartography is changing from an analog to a digital
generation and presentation.
Parallel to the digital development in forest cartography, the
geometric resolution of new generation of satellites is also
increasing. The existing optical satellite systems (TM and
SPOT) did not prove suitable for forest mapping applications in
intensive and small plot forest conditions, as in Middle Europe.
With the relatively low geometric resolution of the existing
satellite systems, the complex conditions of the forest areas can
not be mapped. Consequently, satellite data have not been
implemented in forest map generation.
With the availability of high resolution satellite data with up to
6m in panchromatic band, the use of satellite data in the forest
map preparation chain may be possible with the required map
accuracy.
The present research project on one hand verifies the potential
of high resolution satellite data for digital forest map production
and on the other demonstrates the application of sensor fusion
techniques to combine the spectral and spatial information of
satellite data.
