Winter 2011 GEOG457/657: Lab 2 – Image Fusion
This topic will be discussed further in Tuesday’s lecture – back in 5-158
With increasing volumes of image data, it is becoming common to merge or fuse multispectral data using a panchromatic higher resolution image to create a multispectral (MS) image with the resolution of the ‘intensity’ image. Ideally these are from the same dataset – captured simultaneously. This is undergone both for a ‘sharper display’ and potentially a better classification (and other image processing steps).
In PCI, these options are mostly under
Tools -> Algorithm Librarian-> Image Processing -> Data fusion
Options: FUSE FUSEPCT PCTFUS RGBFUS
But there is also IMGFUSE in there (somewhere - you can search for it to locate it under the all algorithms folder)
General principles:
a. Keep subsets moderate creating fuse images within a screen display (<1000 pixels)
b. Run all operations to RGB viewer first until you get the results you want
c. Capture all good images as indicated for a report document (7-8 are listed)
Lab Components:
1. Subset SPOT PAN and MS to create RGB higher resolution image
2. FUSE and then IMGFUSE on the subset window
3. Use sample from 1999 ETM+ scene to fuse pseudocolor thermal layer with PAN
4. Subset Ikonos data to an area of your choosing and fuse
5. Optional - merge an overlapping subset of SPOT MS and Ikonos PAN
Data : All data are in /home/labs/geog457
a. SPOT 10m PAN and 20m MS from the same Image Date: 6 Sept 2008
pgspot8sept06pan.pix and pgspot8sept06ms.pix
This image (from geobase.ca) is mostly south of PG but includes half the city including the area around UNBC.
b. Ikonos 2002 – Cranbrook Hill
1m PAN: pgikon1m.pix 4m MS: pgikon4m.pix
Data are 11-bit (Range 0- 2047)
c. ETM+ images from 12 Sept 1999:
Pan (channel 8) – 15m res
MS bands 1-5, 7 - 30m res
Thermal channel 6 (low gain) tm61 Thermal channel 6 (high gain) tm62 (60 m res)
Steps:
1. SUBSET (CLIP) SPOT IMAGES
Create from the SPOT image a subset of your choice – you should recall the steps from Thursday tutorial. We do not have to perform any re-projections on these data so you may want to decide on a nice clean extent from the first clip (subset).
-Why do we decide this for the first subset?
- Why do we only need one subset?
- Which image should you subset first?
Keep your subset extent (X or Y) to no more than 500 pixels (MS) or 1000 (PAN) .. [~10km] - otherwise you always need to zoom to see the effect of fusing. In fact, for the sake of practice, subset at exactly 500 (MS) or 1000 (PAN) pixels.
- What is the projection ?
- How did we determine this?
- How did you get the exact number of pixels - were the extents nice and clean?
2A. FUSE
Input- bands 4,3,2 MS for best contrast; Parameters- nearest / cylinder recommended.
- What does loading the band combination of 4,3,2 in focus remind you of (think TM)?
Run just to display first to test, then to a new file when happy -
- How do we save this new file?
- What should we name it?
- We should add some information for channel descriptions - how to we do this?
- How can we add other metadata?
Output
Use can use the GIMP to screen grab the before, and FUSE images as a .png file to be saved in an open office (or AbiWord) document. You can also get better quality images by zooming to the area you are interested in, printing the screen and then opening up the file in GIMP. Scott will show you both ways.
2B. Using fused images in classification
Step 1 just creates a sharper image – can we use it in image processing? Briefly try an unsupervised classification (input bands fused 4-3-2). For specific classes e.g. small lakes - Roger suggests 10-12 classes.
- Do you remember how to perform an unsupervised classification?
- What do you need to do to the PCI file first before the classification?
- Is the higher res useful ? …It may be enough to try on the new data without having to repeat on the 20 m data.
2C. IMGFUSE
Like FUSE this includes the HIS and RGB transformations, but IMGFUSE appears to preserve the initial radiometric values (DNs), so it should be better for classification and other processes. Run it on the same subset and visually compare with the FUSE results. (save the new channels to the same file you used in FUSE above)
It seems parameters InputLock and TransfText can be defaulted as the SPOT data are cleanly co-registered. The help suggests that if both images are geo-rectified, there is no need for InputLock and TransfText. As mentioned our images are not only geo-rectified, but they are also have their pixels lining up properly (the same as our images on Thursday).
i. Start with parameter defaults: maxgain = 3, kernel length=7. The help suggests reducing gain if there is noise in the water, and max value should not exceed 7.
ii. Also run this using 7 and 3 instead of 3 and 7 and check the differences. Save the better one.
- What are InputLock and TransfText input ports - see Roger's notes below - LOOK AT THE HELP
- What is an input port?
- How did we save the channels created in the IMFUSE to new channels in the FUSE file
- How many channels should there be in this file?
- What should we do, once we have all the channels we want in the file - and how?
Output
Use GIMP or the high quality printing to capture the IMGFUSE (4-3-2) image
iii. unsupervised classification – input bands 4-3-2
Compare DNs for features such as water, cutblock and forest with the original bands, and the FUSE result.
Output
Produce any additional noteworthy image results
Note down issues and results for class on image merging on tuesday
IMGFUSE reference Overview (from ‘help’)
Normally the IMGFUSE procedure is run after running IMGLOCK to lock the low, and high resolution images. The result of the IMGLOCK process is a precision corrected low resolution copy of the high resolution image; the image also contains transformation information in a text layer (TransfText) indicating the resolution between the low and high resolution images.
If the low resolution (InputIma) and high resolution (InputRef) images are already geocoded products, the image locking process (IMGLOCK) can be omitted, and IMGFUSE can be run directly. In this case the InputLock and TransfText input layers can be defaulted, and the required information will be derived from the geocoded images. For this process to work, the geocoded images must be in the same georeferencing system; they cannot be in different projections.
3. FUSEPCT
Use the ETM+ subset data to try the FUSEPT option – simulating a high resolution pseudocolor by merging thermal band PCT (60m) with PAN (15m).
First copy the 3 PCI (pix) files from the lab directory to your workspace. Then open the images into focus.
- Did you create a new directory for this work?
- How did you know which files to copy?
Display thermal band 6 as pseudocolor, first low gain (61) then high gain (62). Examine their histograms to see the difference – which has the higher average DN and standard deviation?
Initially the displays will be low contrast as DNs are low range. Change this by right-clicking on map name, and select editPCT, and change min X and max X to approximate the histogram (first for 61, then later 62) then click ‘compress’ and if happy with the results, save this PCT. Use this with the initial thermal band as inputs to FUSEPCT
- What were the steps you took to setup your PCT - what about the low gain image?
- What was the best presentation from the colour selection panel?
Repeat for thermal band 62 and compare results
Output
Create images of the two fused images for the record
4. IKONOS DATA
You can either finish the subset of the Ikonos 4 metre MS and 1m PAN images you started last Thursday, or you can just clip out two images from the original datasets (the second method does not require re-projecting. The re-projecting seems to screw up the data somehow. – they should be <1000 pixels.
Run FUSE – using bands 3-2-1 (visible wavelengths).
- Are the results similar to those with SPOT apart from the wavelengths?
Create images the two images (Before and FUSE)
IMGFUSE
This does not appear to run with IKONOS data .. perhaps because it is 11-bit ?
Both files can be converted (one at a time) to 8 bit using SCALE (Algorithm librarian)
SCALE the subset Ikonos data using these parameters - Scott will illustrate
Input: band 1 for the 1 metre PAN and subsequently bands 3-2-1 MS
Tail trimming (left and right): 1.5
Scaling function: LIN
Output type: 8 bit unsigned
Run first to Grayscale Viewer - RGB and check results
If you can subsequently fuse the now 8 bit layers, GIMP the result
5. Optional – untested …
How well can we merge different dates – ETM: 1999 Ikonos: 2002 SPOT 2008.
The ikonos has no mid-IR band, but could we simulate one at high-res by fusing SPOT432 with the Ikonos PAN ?
SPOT and ETM are the same season, so no major seasonal differences between these two
Use a subset to merge either Ikonos or ETM PAN with SPOT MS to see how well we can generate high res mid IR, and/or what happens between different datasets.
There will be oddities where things have changed between dates – anything else?
Related PCI operations not used in this lab:
IHS: used within FUSE
PCTFUS: difference with FUSEPCT ?
RGBFUS: difference with FUSE?