1. Image fusion (enhancement technique)
We will learn a simple method for fusion (merging/combining) high spatial resolution images with multi-spectral imagery. This fusion technique is combines the higher spatial resolution (15 m pixel size) of band8-Panchromatic with multi-spectral images (30 m) of Landsat 7 ETM+.
A. Firstly, enhance the band8 with a high pass filter. To do this, select the following menu items:
Choose Filter type: Linear, Filter name: Edgesenh. Type output name : B8filter. and click define (not to show the result, only execute the algorithm)
B. Stretch the filtered image. To do this step, select the following menu items:
Select B8filter as input map. Choose Linear Stretching and give a Percentage of 1. Type output name: B8stretch.
C. Mask the image B8mask excluding water (darks pixels) from the input range. The output name will be B8mask.
Type the following formula on the command line of the main ILWIS window:
B8mask=iff((B8stretch>225)or(B8stretch<50),?,B8stretch) , click enter and select domain “Image”.
D. Increase the spatial resolution of ETM+ bands 7, 5 and 3.
To do this, select Operations-Image Processing-Resample. Specify as georeference: band8 (this assures that the other ETM+ bands are densified to the dimensions and pixel size of the band8-panchromatic).
Select a Bilinear resampling algorithm. Call the output maps: B3, B5 and B7. Click the check box Show.
E. The enhanced B8mask can be merged by normalizing the TM bands by their sum and multiplying with B8mask
Calculate three new bands containing the merged images as follows:
R=(B7/(B3+B5+B7))*B8mask ,hit enter
G=(B5/(B3+B5+B7))*B8mask ,hit enter
B=(B3/(B3+B5+B7))*B8mask ,hit enter
F. Create a color composite of the red, green and blue channels, type output name: RGB_fusion, then compare this image with the color composite of the original TM bands. Make sure again that all images are assigned an “Image” domain, as the color composite operation currently only accepts the domain: Image.
Interpret the differences between RGB fusion image and RGB original image