Practical 2 Preprocessing
Uploaded by
manojkbhat61Practical 2 Preprocessing
Uploaded by
manojkbhat61Remote Sensing
Practical 2: Data-Preprocessing
Aims and Objectives:
The aim of this practical is to introduce some of the data-preprocessing technique for remote
sensing image in ENVI environment.
Core Task:
1. Layer Stacking
2. Image Sub-setting
3. Image Mosaicking
4. Image Sharpening
At the end of this practical, you should be familiar with some of the data-preprocessing techniques
for the remote sensing in ENVI.
Task 1: Layer Stacking
Introduction to the data
The image file you will be working with this is a Landsat Enhanced Thematic Mapper (ETM)
image of part of Nepal. The file in in zipped format and contains the 7 Landsat and ETM bands.
Information Box
Layer stacking is often used to combine separate image bands into a single multispectral image
file. Layer stacking is also commonly used to combine image derivatives with spectral bands for
further analysis (i.e. layer stack an NDVI image with spectral bands for input to an image
classification). In order to do that the images should have the same extent (number of rows and
number of columns), which means you will need to resample other bands which have different
spatial resolution to the target resolution. In other words, all images/bands should have same
spatial resolution to be able to perform layer stacking. However, combining images/bands will
increase the final stacked image size, and consequently will increase the processing time later when
you do your analysis.
1. Copy zipped image data from image stacking folder to your working folder and unzip it.
After unzipping, you can see 7 separate ETM band of Landsat imagery.
2. Add all image file to the ENVI except band 6, which is thermal band.
3. In the toolbox at the right side of ENVI interface, click on raster management > layer
stacking. Or, you can search as ‘Layer Stacking” at the top of toolbox. The layer stacking
parameters dialog box appears.
Figure 1: Layer stacking parameters dialog box
4. Click on import file and you can see new dialog box naming Layer Stacking Input file.
Figure 2: Layer stacking input file dialog box
5. In this dialog box, all the image band appears in the select input file. Be sure that all the
image bands are in order i.e. band 1 at first and band 7 at last. If not, make it sure that all
are in order. If everything is ok, select all the band and click ok.
6. In output file range, click on inclusive: range encompasses all the files.
7. In output result, choose file option and browse to your working directory to save the
8. In right side of the dialog box, there should appear the original projection system of image
(UTM- WGS 84, Zone 45N) and pixel size of 30 m. If not, make sure that correct projection
system is filled. In Resampling, Choose Nearest Neighbor resampling technique.
9. Click ok to proceed the layer stacking.
10. After finishing layer stacking, you can visualize the layer stacked image in natural colour
or false colour composite.
Question 1:
What is the advantage of doing layer stacking rather than leaving each band as individual?
Task 2: Image Sub-setting
Image sub-setting is the process by which a subset of the raster dataset is created. Clipping removes
data outside the area of interest reducing the file size and improving the processing time for many
operations.
Data Used:
• Layer stacked image
• Shapefile of Kathmandu Valley (ktmwgs.shp)
1. Close all the data and open the layer stacked image. In the same window, open the shapefile
of Kathmandu valley. You can see at the image window that vector file of Kathmandu
valley is overlapped on the layer stacked image. If not, right click on the shapefile
(ktmwgs.shp) and click on zoom to layer extent.
2. Check the metadata of the image and shapefile and be sure that both of them are in same
projection system.
3. In the toolbox, click on region of interest > subset data from ROIs. The select input file
to subset via ROIs dialog box appears.
Figure 3: Subset via ROI dialog box
4. In the select input file, click on layer stacked image and click ok.
5. Then, spatial subset via ROI parameters dialog box appears.
Figure 4: Spatial subset via ROI dialog box
6. In select input ROIs, select the ktmwgs.shp file.
7. In mask pixels output of ROI, click on the double arrow at right side to appear mask
background value and put 0 in the box.
8. Choose your working directory and place suitable name and click ok.
Task 3: Image Mosaicking
In many cases, it may not be possible to capture the complete image of a study area in a single
exposure as most of the image capturing detector or sensor work on definite field of view. In such
cases, the document has to be scanned part by part producing split images. Thus, the image analysis
and processing require mosaicking of the split images to obtain a complete final image of the
document.
Mosaicking is one of the techniques of image processing which is useful for tiling digital images.
Mosaicking is blending together of several arbitrarily shaped images to form one large
radiometrically balanced image so that the boundaries between the original images are not seen.
Any number of geocoded images can be blended together along the user-specified cut lines
(polygons).
Data Used:
• Wasia1_mss.img, Wasia2_mss.img, Wasia3_mss.img
Method 1: Seamless mosaic
1. Copy all of the data from the image mosaicking folder to your working folder.
2. Add all three images from wasia folder in the ENVI window.
3. Click Mosaicking > Seamless Mosaic. The seamless mosaic dialog box appears.
4. Click Add ribbon, the file selection dialog box appears. Click on select all > ok. After
this, you can see three images in the scene name.
Figure 5: Seamless Mosaic dialog box
5. Right click on the data ignore value and click on change selected parameters. Put 0
value in the dialog box and click Ok. Now you can see in the image that black
background has changed into the white background.
6. In colour matching action, choose one image which act as reference image as reference
and other as the adjust.
7. Click on recalculate footprints .
8. Click seamlines and click on auto generated seamlines.
9. Click on colour correction > Histogram matching and tick on entire scene.
10. Click on feathering and choose seamline feathering.
11. Click on export. Chose output format as ENVI. Browse on output filename and
navigate towards your working folder and put the output name as mosaicwasia.
12. Chose resampling method as Nearest neighbor.
13. Choose all band as select output bands and click finish.
14. View the mosaic image in suitable colour composite.
Method 2: Pixel Based Mosaicking
1 Click on mosaicking > pixel-based mosaicking in the toolbox. The pixel-based mosaic
dialog box appears.
2 Click on import > import files. The mosaic input files dialog box appears. In select input
file, select three images and click ok.
3 Click on file > Apply to open mosaic parameter dialog box. Check on output result to file
and browse to your working directory and select name as mosaic. Put background image
as 0 and click on ok.
4 View the mosaic image in suitable colour composite.
Question 2:
In the final output from seamless mosaic and pixel-based mosaicking, what difference did
you notice? What is the reason behind it?
Task 4: Image Sharpening
For theoretical concept, please refer two literature in the image sharpening folder.
Data Used:
• High_resolution.img
• Multispectral.img
1 Copy the resolution merging folder to your working folder.
2 Add the two images in ENVI interface, the multispectral and panchromatic images.
Visualize the multispectral image in colour composite.
3 Take a swipe tool to visualize the difference between panchromatic and multispectral
image.
4 In toolbox, click on Image sharpening > PC spectral Sharpening. The select low spatial
resolution multi band input file appears. Select multispectral image here and click ok.
5 This is followed by the select high spatial resolution input file. Chose high resolution image
here and click ok.
6 Select Nearest neighbor resampling technique.
7 Navigate to your working folder and give the name pan sharpen and click ok.
8 Remove all the image in the layer manager. Then add pan sharpen and panchromatic image
and see the difference using swipe tool. Similarly, add pan sharpen and multispectral image
and see the difference.
Exercise: Experiment with other available algorithm in ENVI for the image sharpening.
Question 3:
What is the difference between panchromatic, multispectral and pan-sharpened image?
Question 4:
What do you see the difference in the output of different image sharpening techniques?
Why?
Question 5:
Which resampling technique is suitable for layer stacking, mosaicking and image
sharpening?
