Goal of the lab: To introduce the class to important analytical processes in remote sensing. The lab explores image mosaic, spatial and spectral image enhancement, band ratio and Binary change detection.
Part 1: Image Mosaicking
A image mosaic procedure is performed when an area of study is larger than one satellite image scene or if the area of study is at the intersection on two or more photos. For the purpose of this lab the images being used are from Landsat TM captured in May of 2005 and are in path 25 row 29 and path 26 row 29.
The first step in this process is loading the first image into Erdas and then clicking Multiple tab in the Select Layers To Add window. After that the next step is to chose Multiple Images in Virtual Mosaic. Once that is done go into the Raster Tab, still in the Select Layers To Add window, and make sure Background Transparent is checked. Also make sure Fit to Frame is clicked and then click OK to load the image into the viewer.
The above process should be repeated for the other image to bring it into the viewer. The viewer should now show both images overlapping. Now to Mosaic the images, and there are 2 ways to do this. The first was is using Mosaic Express. In the Raster tab you will find the Mosaic tab and click Mosaic Express. Once that window comes up click the folder to add the two images. Make sure that the image wanted on top is added first. Clicking next will take you to the Input Area Dialog, for this process it is not necessary so keep clicking next until the Output Dialog tab shows up. This is where the image will be saved. Then accept the default parameters and click Finish to run the model.
Part 1: Image Mosaicking
A image mosaic procedure is performed when an area of study is larger than one satellite image scene or if the area of study is at the intersection on two or more photos. For the purpose of this lab the images being used are from Landsat TM captured in May of 2005 and are in path 25 row 29 and path 26 row 29.
The first step in this process is loading the first image into Erdas and then clicking Multiple tab in the Select Layers To Add window. After that the next step is to chose Multiple Images in Virtual Mosaic. Once that is done go into the Raster Tab, still in the Select Layers To Add window, and make sure Background Transparent is checked. Also make sure Fit to Frame is clicked and then click OK to load the image into the viewer.
The above process should be repeated for the other image to bring it into the viewer. The viewer should now show both images overlapping. Now to Mosaic the images, and there are 2 ways to do this. The first was is using Mosaic Express. In the Raster tab you will find the Mosaic tab and click Mosaic Express. Once that window comes up click the folder to add the two images. Make sure that the image wanted on top is added first. Clicking next will take you to the Input Area Dialog, for this process it is not necessary so keep clicking next until the Output Dialog tab shows up. This is where the image will be saved. Then accept the default parameters and click Finish to run the model.
The images have now been through the Mosaic process but you can see that the colors are different and it is very easy to tell where the border is.
MosaicPro is another tool to Mosaic images together. The begging of this process is the same as Mosaic Express, including the way to being the images into the viewer. When this is done Click Mosaic and then MosaicPro. When this window opens Add Images is the first click to add the images you want to mosaic. Highlight the first image to be added and click Image Area Options tab and then Compute Active Area Options. This activates the Set button which should now be clicked. Then click Ok in the Add Image window. Repeat this process for the second image to be added.
To make the color difference less so, the radiometric properties at the overlap can be synchronized. To do this click Color Corrections and then Use Histogram Matching and click Set. In the new dialog window select Overlap Areas as the Matching Method. Then click OK in the Histogram Matching dialog and the Color Corrections dialog.
Back in the MosaicPro toolbar click on the Set Overlap Functions icon. In the new window make sure Overlay is the method being used. Once this is done click okay. To run the model click Process then Run Mosaic.
MosaicPro is another tool to Mosaic images together. The begging of this process is the same as Mosaic Express, including the way to being the images into the viewer. When this is done Click Mosaic and then MosaicPro. When this window opens Add Images is the first click to add the images you want to mosaic. Highlight the first image to be added and click Image Area Options tab and then Compute Active Area Options. This activates the Set button which should now be clicked. Then click Ok in the Add Image window. Repeat this process for the second image to be added.
To make the color difference less so, the radiometric properties at the overlap can be synchronized. To do this click Color Corrections and then Use Histogram Matching and click Set. In the new dialog window select Overlap Areas as the Matching Method. Then click OK in the Histogram Matching dialog and the Color Corrections dialog.
Back in the MosaicPro toolbar click on the Set Overlap Functions icon. In the new window make sure Overlay is the method being used. Once this is done click okay. To run the model click Process then Run Mosaic.
As you can see this image has much cleaner color change and it looks more like one image. It's much harder to tell where the borders are.
Part 2: Band Ratioing
Performing Band Ratio by implementing NDVI on the image. NDIV being Normalixed Difference Vegetation Index.
Bring in the image to be changed. Then under Raster tools go under Unsupervised then NDVI. Now the Indices interface should be open, this is where the chosen image should be inputed. In the Indices window make sure the Sensor reads Landsat TM, and under Select Function highlight NDVI.
Then click OK to create the image and run the model.
"What will you expect to find in areas that are very white in the NDVI image?" -- Vegetation
"Comment on the presence or absence of vegetation in areas that are medium gray and black." - These are they areas with little to no vegetation.
Part 3: Spatial and Spectral Image Enchancement
This section of the lab is an introduction to spatial enhancement techniques.
"What is a high frequency image?" -- An image with sharp contrast between colors and clear edges and borders.
The image to be worked with is a high frequency image and the plan is to perform a Low Pass Convolution Filter on the image, to do this the tool needed is under Raster tools. Click Spatial then Convolution. Under the Kernel selection menu, select 5x5 Low Pass. Input the image needed and name and pick a place to save it. The rest of the parameters should be accepted and hit OK. If the images are then compared the differences should be clear.
"Outline the differences between the original image and the 5x5 Low Pass filtered image you
just created." -- The new images is a little bit blurrier and looks a little bit bigger.
The next part of the lab is taking a low frequency image and improving it with a high pass filter.
A new image was brought into Erdas for this.
"What is a low frequency image?" -- An image that does not have high contrast and things blur together and look smooth.
To improve the low contrast image a 5x5 High Pass Convolution Filter will be applied in a similar manner as the Low Pass Filter, the difference being choosing the High Pass Filter in the Kernel Selection option.
"Outline the differences between the original image and the 5x5 High Pass filtered image
you just created." -- The image is much darker but more well defined and clearer lines.
Edge Enhancement
A new image was brought in for this part. for this process the Convolution tool bar is to be accessed. The Kernel this time will be 3x3 Laplacian Edge Detection. Under the Handle Edges By option click Fill. Uncheck the Normalize the Kernel option, then click OK.
"What is a Laplacian convolution filter?" -- A filter that highlights regions of rapid intensity change.
"Outline the differences between the original image and the Laplacian edge detection image
you just created." -- The second image is a lot darker, but rivers are much more noticeable. There is also a slight criss-cross pattern across the picture.
Part 2: Band Ratioing
Performing Band Ratio by implementing NDVI on the image. NDIV being Normalixed Difference Vegetation Index.
Bring in the image to be changed. Then under Raster tools go under Unsupervised then NDVI. Now the Indices interface should be open, this is where the chosen image should be inputed. In the Indices window make sure the Sensor reads Landsat TM, and under Select Function highlight NDVI.
Then click OK to create the image and run the model.
"What will you expect to find in areas that are very white in the NDVI image?" -- Vegetation
"Comment on the presence or absence of vegetation in areas that are medium gray and black." - These are they areas with little to no vegetation.
Part 3: Spatial and Spectral Image Enchancement
This section of the lab is an introduction to spatial enhancement techniques.
"What is a high frequency image?" -- An image with sharp contrast between colors and clear edges and borders.
The image to be worked with is a high frequency image and the plan is to perform a Low Pass Convolution Filter on the image, to do this the tool needed is under Raster tools. Click Spatial then Convolution. Under the Kernel selection menu, select 5x5 Low Pass. Input the image needed and name and pick a place to save it. The rest of the parameters should be accepted and hit OK. If the images are then compared the differences should be clear.
"Outline the differences between the original image and the 5x5 Low Pass filtered image you
just created." -- The new images is a little bit blurrier and looks a little bit bigger.
The next part of the lab is taking a low frequency image and improving it with a high pass filter.
A new image was brought into Erdas for this.
"What is a low frequency image?" -- An image that does not have high contrast and things blur together and look smooth.
To improve the low contrast image a 5x5 High Pass Convolution Filter will be applied in a similar manner as the Low Pass Filter, the difference being choosing the High Pass Filter in the Kernel Selection option.
"Outline the differences between the original image and the 5x5 High Pass filtered image
you just created." -- The image is much darker but more well defined and clearer lines.
Edge Enhancement
A new image was brought in for this part. for this process the Convolution tool bar is to be accessed. The Kernel this time will be 3x3 Laplacian Edge Detection. Under the Handle Edges By option click Fill. Uncheck the Normalize the Kernel option, then click OK.
"What is a Laplacian convolution filter?" -- A filter that highlights regions of rapid intensity change.
"Outline the differences between the original image and the Laplacian edge detection image
you just created." -- The second image is a lot darker, but rivers are much more noticeable. There is also a slight criss-cross pattern across the picture.