Image fusion

From AWF-Wiki
(Difference between revisions)
Jump to: navigation, search
(Create a synthesis panchromatic band)
 
(17 intermediate revisions by 2 users not shown)
Line 1: Line 1:
Pansharpening of Sentinel-2 requires defining a 10m reference band because there is no panchromatic band as for Landsat available. The only option is a selection among the 10-m bands (2, 3, 4, or 8) or synthesis (e.g. weighted average of 10-m band). Following the latter approach, the luminance can be calculated from the Blue, Red Green and NIR band to receive a single band grey scale image.
+
Pansharpening of Sentinel-2 requires defining a 10m reference band because there is no panchromatic band as for Landsat available. One option is a selection among the 10-m bands (B2, B3, B4 or B8) or a synthesis (e.g. weighted average of 10-m bands). Following the latter approach, luminance can be calculated from Blue, Red, Green and NIR band to receive a single band grey scale image.
  
 
= Create a synthesis panchromatic band =
 
= Create a synthesis panchromatic band =
 
# Load the input image
 
# Load the input image
# Open the raster calculator and apply the following formula: Y = 0.20 * NIR(B8) 0.15 * Red(B4) + 0.58 * Green(B3) + 0.07 * Blue(B1)
+
# Open the raster calculator and apply the following luminance formula: Y = 0.20 * NIR 0.15 * Red + 0.58 * Green + 0.07 * Blue
   [[File:rastercalc.png]]
+
   [[File:rastercalc.png|600px]]
 
# Define path and name for the output layer
 
# Define path and name for the output layer
 
# Click on the ‘OK-button’.
 
# Click on the ‘OK-button’.
  
 
= Create a pan-sharpened layer =
 
= Create a pan-sharpened layer =
 +
Does currently not work in Plugin for QGIS 3.4. Use OTB standalone.
 
# Search for Pansharpening (Bayes) algorithm in the Processing Toolbox.
 
# Search for Pansharpening (Bayes) algorithm in the Processing Toolbox.
 
# Select pseudo-panchromatic band as the Input PAN Image.
 
# Select pseudo-panchromatic band as the Input PAN Image.
# Select the layer contain the original 20m bands as the InputXS Image.
+
# Select the layer stack with the original 20m bands as the InputXS Image. Therefore  (1) Follow [[Split stack]] to extract the  20m Bands ('''additional creation parameters: ''':{{typed|text=-b 5 -b 6 -b 7 -b 9 -b 11 -b 12}}) from the multiband file creating at the same time a new multiband file.
 +
 
 
# Specify the directory to save the output file and click on the ‘Run-button’.
 
# Specify the directory to save the output file and click on the ‘Run-button’.
  

Latest revision as of 15:52, 17 November 2018

Pansharpening of Sentinel-2 requires defining a 10m reference band because there is no panchromatic band as for Landsat available. One option is a selection among the 10-m bands (B2, B3, B4 or B8) or a synthesis (e.g. weighted average of 10-m bands). Following the latter approach, luminance can be calculated from Blue, Red, Green and NIR band to receive a single band grey scale image.

[edit] Create a synthesis panchromatic band

  1. Load the input image
  2. Open the raster calculator and apply the following luminance formula: Y = 0.20 * NIR 0.15 * Red + 0.58 * Green + 0.07 * Blue
 Rastercalc.png
  1. Define path and name for the output layer
  2. Click on the ‘OK-button’.

[edit] Create a pan-sharpened layer

Does currently not work in Plugin for QGIS 3.4. Use OTB standalone.

  1. Search for Pansharpening (Bayes) algorithm in the Processing Toolbox.
  2. Select pseudo-panchromatic band as the Input PAN Image.
  3. Select the layer stack with the original 20m bands as the InputXS Image. Therefore (1) Follow Split stack to extract the 20m Bands (additional creation parameters: :-b 5 -b 6 -b 7 -b 9 -b 11 -b 12) from the multiband file creating at the same time a new multiband file.
  1. Specify the directory to save the output file and click on the ‘Run-button’.

[edit] Compare the fused image with the original image

No pansharpen.png Pansharpen.png

Personal tools
Namespaces

Variants
Actions
Navigation
Development
Toolbox
Print/export