The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
1143
obtained by low-pass filter with the pan-band, this study use the
low-pass filter operator of 5*5 of ERDAS8.7.
3.1.3 Brovery Change Fusion Method
Brovery fusion formula as follows:
B
lowD high
MB,
Ills
low,,,
(3)
3.1.5 IHS Transform Fusion Method
At present, IHS transform model mainly contains spherical
transform, cylinder transform, triangular transform and six
pyramid transform. On the image fusion, the differences
between the four models are not obvious, but sphere
transform is the best in compare [9] .
In this study use IHS transform method for the fusion
experiment, and programming in Matlab to achieve the fusion
algorithm.
In the formula, d is the fusion image, n is bands numbers,
J-J MB
denominator denote the summation of the three ETM+ multi-
spectral bands.
3.1.4 The Product Transform Fusion Method
The product fusion method is a simple algebra operation, multiply
the pixel of the two images to obtain the fusion image. In this
paper, take extraction with the product of the pixel to obtain the
fusion image pixel value, the fusion algorithm formula as follows:
ML^^XS^PN^
3.1.6 Principal Component Fusion Method
The image after fusion contains the character of high spatial
resolution image and high spectral resolution of the original
image, reserve the high frequency of the original image. The
individual part character of the aim of the fusion image is
more clear, and the spectral information is more richer, and
PCA transform conquer the localization of the HIS transform
which can only at the same time do fusion with three bands,
it can do fusion with more than two multi-spectral image.
3.2 Fusion Image Evaluation Parameters
The common used image fusion quality evaluation
parameters are: the mean, standard deviation, average
gradient, information entropy, and the correlation coefficient.
In the formula, ML. jk is the fusion image pixel value, XS jjk is the
pixel value of multi-spectral image, PN.. is the pixel value of
panchromatic.
4. EXPERIMENTS AND ANALYSIS
Programming the various fusion algorithm with Erdas Model
module and matlab, fused images are displayed: 5, 4 and 3
bands in accordance with the R, G, B, fusion results are
shown as follows(figurel-6):
Figure4.ML fused image 543bands Figure5.PCA fused image 543bands Figure6. HIS fused image 543bands