In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
383 
MULTIRESOLUTION IMAGE FUSION: PHASE CONGRUENCY 
FOR SPATIAL CONSISTENCY ASSESSMENT 
A. Makarau a ’ , G. Palubinskas a , P. Reinartz a 
a DLR, German Aerospace Center, 82334 Weßling, Germany - (aliaksei.makarau, gintautas.palubinskas, 
peter.reinartz)@dlr.de 
Commission VII, WG VII/6 
KEY WORDS: pan-sharpening, multispectral image fusion, spatial consistency assessment, phase congruency 
ABSTRACT: 
Multiresolution and multispectral image fusion (pan-sharpening) requires proper assessment of spectral consistency but also spatial 
consistency. Many fusion methods resulting in perfect spectral consistency may leak spatial consistency and vice versa, therefore a 
proper assessment of both spectral and spatial consistency is required. Up to now, only a few approaches were proposed for spatial 
consistency assessment using edge map comparison, calculated by gradient-like methods (Sobel or Laplace operators). Since image 
fusion may change intensity and contrast of the objects in the fused image, gradient methods may give disagreeing edge maps of the 
fused and reference (panchromatic) image. Unfortunately, this may lead to wrong conclusions on spatial consistency. In this paper 
we propose to use phase congruency for spatial consistency assessment. This measure is invariant to intensity and contrast change 
and allows to assess spatial consistency of fused image in multiscale way. Several assessment tests on IKONOS data allowed to 
compare known assessment measures and the measure based on phase congruency. It is shown that phase congruency measure has 
common trend with other widely used assessment measures and allows to obtain confident assessment of spatial consistency. 
1. INTRODUCTION 
Pan-sharpened data have many areas of application and 
therefore different requirements are posed on the fusion 
method. The requirements can be on spectral consistency, 
spatial consistency or on the both together. Spectral consistency 
assumes that pansharpened data have increased spatial 
resolution with spectral properties of the original data. Spatial 
consistency assumes that “A high spatial quality merged image 
is that which incorporates the spatial detail features present in 
the panchromatic image and missing in the initial multispectral 
one” (Gonzalez-Audicana, 2005). Spectral and spatial 
consistency together is the ideal case of the fused data and the 
ideal fusion method is to provide these characteristics. A pan- 
sharpening method may provide perfect spectral consistency 
together with poor spatial consistency and vice versa. 
Therefore, to make proper decision on a fusion method (or to 
outline the best one), assessment of both spectral and spatial 
consistency is to be performed. 
<9(r,f) = cos 
-l 
If=i r ifi 
ifcUj 2 i£i// 2 
(i) 
K is the number of bands, r and f are the two vectors created by 
the values of spectral channels at the same pixel in the reference 
and fused images A and 5; Structural SIMilarity SSIM (Wang, 
2004) or extended SSIM - Q4 (Alparone, 2003), (correlation, 
contrast, and luminance similarity between two images are used 
to calculate one similarity value): 
SSIM(A i ,B i ) = 
( 0 A;B: +C 3 A 
2 MAi^Bi +c \ 
2 a A i cr B i +^2 
v s Ai + s Bi y 
v +c 2 ^ 
& A, < Ts y + C 3 
(2) 
2. PAN-SHARPENED DATA QUALITY 
2.1 Spectral consistency 
Spectral consistency assessment usually performed using 
Wald's protocol in order to make reference multispectral data of 
high resolution available. There is a variety of developed and 
well-known similarity measures used for spectral consistency 
assessment. The most known and popular are: Spectral Angle 
Mapper, SAM (calculated as the angle between two vectors): 
where ju A and /j b are the local sample means of Aj and B 
respectively, a A and cr 5 . are the local sample standard 
deviations of A j and Bj, respectively, and cr A , B , is the sample 
cross correlation of Aj and Bj after removing their means. The 
items C}, C2, and C3 are small positive constants that stabilize 
each term; ERGAS (Wald, 1997) (similarity measure for 
multispectral images, based on MSE estimator): 
ERGAS(A i ,B i ) = 100y 
1 ^RMSEjAj^j) 2 * 
i K ^ Sa, 
(3) 
* Corresponding author. Tel. +49-8153-283672