683 
In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
scheme. To improve the spectral quality, the high-pass 
details are injected proportionally to the low-pass MS 
components in such a way that the fused MS pixel vector is 
always proportional to that before fusion. Aiazzi et al present 
the GLP-CBD fusion algorithm (Aiazzi, et al 2002), which 
exploits MRA, achieved through GLP, with the spatial 
frequency response of the analysis filters matching a model 
of the modulation transfer function (MTF) of the MS 
instrument. The injection model employs a decision based on 
locally thresholding the correlation coefficient (CC) between 
the resampled MS band and the low pass approximation of 
the Pan. Ranchin et al. (Ranchin et al, 2003) present the 
“Amélioration de la Résolution Spatiale par Injection de 
Structures” (ARSIS, Improving Spatial Resolution by 
Structure Injection) concept based on the assumption that the 
missing information is linked to the high frequencies of the 
datasets to be fused. Some fusion techniques jointly using 
component substitution with multi-scale analysis were 
developed, such as the algorithms combing wavelet 
transform and IHS transform (Gonzâlez-Audicana, et al, 
2004, Chibani and Houacine, 2002, Zhang and Hong, 2005) 
or PCA transform (Gonzâlez-Audicana, et al, 2004, Shah, 
Younan, and King, 2008). These hybrid schemes use 
wavelets to extract the detail information from one image 
and standard image transformations to inject it into another 
image, or propose improvements in the method of injecting 
information (e.g. Garzelli and Nencini, 2005; Otazu et al., 
2005). Otazu et al introduce sensors’ spectral response and 
ground spectral features into fusion technology on the basis 
of MRA (Otazu, et al, 2005). 
Other authors utilize the regularization method to optimize 
the fusion results so as to satisfy the higher resolution 
multispectral image model (Aanaes, et al. 2008). Yang, et al. 
(2009) generalized this idea and proposed a new model 
quantifying the mathematical relationship between the fused 
higher multispectral images and the original multispectral 
image, the spatial details being extracted from the 
high-resolution panchromatic image, and the adopted fusion 
strategies. 
5. DISCUSSIONS AND CONCLUSIONS 
Currently, the pixel-level image fusion algorithms are 
divided into three categories, i.e., CS technique, modulation 
based technique and MRA based technique according to 
fusion mechanism. With these three categories, similarity 
and difference between fusion techniques can be derived, 
which is important for applications. We discuss two typical 
classes of fusion application, i.e., automatic classification 
and visual interpretation. Automatic classification relies on 
the spectral feature than spatial details, while visual 
interpretation is opposite. Thus, if the fused images are used 
for automatic classification, modulation based technique and 
MRA technique with a lower number of decomposition 
levels are preferable, which better preserve the spectral 
characteristics of multispectral bands. For visual 
interpretation, which benefits from spatial and textural 
details, CS technique and MRA technique with a higher 
number of decomposition levels are appropriate.