In: Wagner W., Szekely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
680 
PIXEL LEVEL FUSION METHODS FOR REMOTE SENSING IMAGES: A CURRENT 
REVIEW 
Yang Jinghui*, Zhang Jixian, Li Haitao, Sun Yushan, Pu Pengxian 
Chinese Academy of Surveying and Mapping, Lianhuachi Xi Road 28, Beijing 100830, P. R. China 
*: Corresponding author. Email: jhyang@casm.ac.cn. Tel: +86-10-63880532. Fax: +86-10-63880535. 
KEYWORDS: Image Fusion, Pansharpening, Pixel Level, Remote Sensing 
ABSTRACT: 
Image fusion is capable of integrating different imagery to produce more information than can be derived from a single sensor. So 
far, many pixel level fusion methods for remote sensing images have been presented, in which the lower resolution multispectral 
image’s structural and textural details are enhanced by adopting the higher resolution panchromatic image corresponding to the 
multispectral image. For this reason, it is also called pansharpening. In this paper we will list current situation of pixel level image 
fusion by dividing those methods into three categories, i.e., component substitution technique, modulation based technique and 
multi-resolution analysis based technique according to fusion mechanism. Also, the properties of the three categories for 
applications are discussed. 
1. INTRODUCTION 
Data fusion is capable of integrating different imagery data 
to produce more information than can that be derived from a 
single sensor. There are at least two limitations accounting 
for demanding pixel level image fusion technology. One is 
that the received energy of multispectral sensor for each 
band is limited because of the narrow wavelength range of 
the multispectral band. In general, the values of Ground 
projected Instantaneous Field Of View (GIFOV) 
(Schowengerdt, 1997) of multispectral bands are larger than 
those of the panchromatic bands. In order to obtain smaller 
GIFOV value in relatively narrow wavelength range, image 
fusion technology is demanded to enhance structural and 
spatial details. The other is that the capability transmitting 
the acquired data to the ground is restricted. However, at 
present the transmission equipments of remote sensing 
system can not address the requirements. Henceforth, after 
ground stations receives multispectral images containing 
relatively less data, the combination of multispectral bands 
with the higher resolution panchromatic band can resolve the 
problem to some extent. So far, many pixel-level fusion 
methods for remote sensing image have been presented 
where the multispectral image’s structural and textural 
details are enhanced by adopting the higher resolution. 
In the recent literature, IEEE Transaction on Geoscience and 
Remote Sensing had published a special issue on data fusion 
in May 2008, which includes several new developments for 
current situation of image fusion (Gamba and Chanussot, 
2008). In January 2006, the Data Fusion Committee of the 
IEEE Geoscience and Remote Sensing Society launched a 
public contest for pansharpening algorithms (Alparone, et al., 
2007), which aimed to identity the ones that perform best. 
The fusion results of eight algorithms (GLP-CBD, AWLP, 
GIHS-GA, WiSpeR, FSRF, UNB-Pansharp, WSIS, 
GIHS-TP) from worldwide participants were assessed both 
visually and quantitatively. These published literatures show 
that data fusion for remote sensing as an active research field 
has attractive interests. This paper will review the current 
situation for pixel-level image fusion technology. 
Typically, the algorithms for remote sensing image pixel 
level fusion can be divided into three general categories 
shown in Fig. 1 (Yang, et al, 2009): component substitution 
(CS) fusion technique (Pellemans et al, 1993; Shettigara, 
1992; Chavez, 1991; Aiazzi, 2007), modulation-based fusion 
technique (Liu, 2000; Zhang, 1999; Gangkofner, et al., 2008) 
and multi-resolution analysis (MRA) based fusion technique 
(Aiazzi, 2002; Amolins, et al., 2007). In addition, some 
fusion techniques integrating component substitution with 
multi-resolution analysis were developed, such as the