International Archives of Photogrammetry and Remote Sensing, Vol. 32, Part 7-4-3 W6, Valladolid, Spain, 3-4 June, 1999 
197 
conditions of each channel, the LCM result looks more realistic. 
The texture appears crisp and sharp at the right place, also 
within areas, where the panchromatic band shows moderate or 
little texture only. Focusing on forest areas, it seems that the 
LCM result represents a superior approximation of the true 
image. 
The characteristic features of the other fusion categories can be 
also observed in Figure 3. The filter techniques, here re 
presented by the HFM result, strictly merge the datasets: the 
(bilinearly resampled) multispectral data appear without 
noticeable distortion; neither positive nor negative correlation 
has been considered. The sharpening effect of the Brovey and 
IHS algorithm looks similar to that of the HFM result, but 
substantial spectral distortions become visible. These 
observations result from the specific properties described in 
chapter 2.1. The multiplication and the IHS method most 
severely distort the spectral information, particularly if the NIR 
channel is included. 
6. CONCLUSIONS 
High resolution and large area covering multispectral data 
would be invaluable for Mid-European forest inventories. 
Today, the fusion of spacebome high resolution panchromatic 
data and low resolution multispectral data appears to be the only 
possibility to obtain such datasets. 
To prevent visual misinterpretations, as well as errors resulting 
from computer-aided analysis, we would like to formulate the 
following requirement for using fusion techniques in forestry: as 
a minimum demand, merging procedures should preserve the 
spectral characteristics of the low resolution multispectral input 
as much as possible. Especially, the original information of the 
important NIR and SWIR channels should not be significantly 
distorted. Therefore, fusion techniques summarised under the 
first category (band-arithmetic, component substitution) are not 
considered adequate for forestry applications, while filter 
techniques, such as the HFM algorithm, comply with this 
minimum requirement. 
In addition, the fusion procedure should optimally relate the 
textural properties of the high resolution panchromatic reference 
to the multispectral information of the (here available) true 
image. This involves local adjustment of the texture 
information. We found that only the LUT and LCM algorithms 
fulfil this requirement. The quality of advanced wavelet 
algorithms (e.g. Iverson and Lersch, 1994; Ranchin et al., 1994; 
Ranchin and Wald, 1996) could not be assessed so far because 
no detailed processing schemes are published. 
The quality of a fusion algorithm should not only be assessed 
regarding the sharpening effect and the radiometric fidelity of 
the result, but also the flexibility, robustness and 
implementation simplicity of the algorithm. We should mention 
that the performance of the LCM approach depends on good 
signal-to-noise ratio and dynamic range, as well as the 
geometric correspondence between multispectral input and 
(degraded) panchromatic band. These prerequisites are satisfied 
in case of the DPA validation dataset. When performing 
multisensor fusion (e.g. fusing TM multispectral data with 
SPOT panchromatic data), the geometry aspect could become a 
limiting factor. Also a low dynamic range (e.g. that of the IRS- 
1C panchromatic band) could lead to badly defined local 
regression models. Therefore, our next investigations will 
concentrate on validation experiments with existing sensor 
configurations. 
7. REFERENCES 
Aplin, P., Atkinson, P.M. and Curran, P.J., 1997. Fine spatial 
resolution satellite sensors for the next decade. International 
Journal of Remote Sensing, 18(18), pp. 3873-3881. 
Carper, W.J., Lillesand, T.M. and Kiefer, R.W., 1990. The use 
of intensity-hue-saturation transformations for merging SPOT 
panchromatic and multispectral image data. Photogrammetric 
Engineering and Remote Sensing, 56(4), pp. 459-467. 
Chavez, P.S., 1984. Digital processing techniques for image 
mapping with Landsat TM and SPOT simulator data. Proc. of 
the 18th International Symposium on Remote Sensing of 
Environment, Paris, 1-5 October, pp. 101-116. 
Chavez, P.S. and Bowell, J.A., 1988. Comparison of the 
spectral information content of Landsat Thematic Mapper and 
SPOT for three different sites in Phoenix, Arizona region. 
Photogrammetric Engineering and Remote Sensing, 54(12), pp. 
1699-1708. 
Colvocoresses, A.P., 1977. Proposed parameters for an 
operational Landsat. Photogrammetric Engineering and Remote 
Sensing, 43(9), pp. 1139-1145. 
Filiberti, D.P., Marsh, S.E. and Schowengerdt, R.A., 1994. 
Synthesis of imagery with high spatial and spectral resolution 
from multiple image sources. Optical Engineering, 33(8), pp. 
2520-2528. 
Fritz, L.W., 1996. The era of commercial earth observation 
satellites. Photogrammetric Engineering and Remote Sensing, 
62(1), pp. 39-45. 
Garguet-Duport, B., Girel, J., Chassery, J.-M. and Pautou, G., 
1996. The use of multiresolution analysis and wavelets 
transform for merging SPOT panchromatic and multispectral 
image data. Photogrammetric Engineering and Remote Sensing, 
62(9), pp. 1057-1066. 
Hallada, W.A. and Cox, S., 1983. Image sharpening for mixed 
spatial and spectral resolution satellite systems. Proc. of the 
17th International Symposium on Remote Sensing of 
Environment, 9-13 May, pp. 1023-1032. 
Haydn, R., Dalke, G.W. and Henkel, J., 1982. Application of 
the IHS color transform to the processing of multisensor data