Before completing lab 4 a general understanding of Edras Imagine and its functions is necessary. Through the first 3 exercises this semester Professor Cyril Wilson led classroom discussions and hands on in class tutorials of how to complete various tasks. This lab is designed to further skills in a number of different areas. These skills include delineating a study area from a larger satellite image, showing how changing spatial resolution can better the uses for visual interpretation and becoming familiar with radiometric enhancement techniques. From there it is also important to be able to link Google Earth with Erdas to utilize the benefits of both. Additionally, resampling , image mosaicking and binary change detection methods were also learned throughout this lab. The purpose of this lab is to gain skills in all of the areas discussed above.
Methods
Part 1
The first step to this lab is to open Erdas Imagine and then to set up a designated area to save the project that has a specific title including the users last name so it will not be mixed up. From there the image eau_claire_2011.img was brought into Erdas, it is important to note that all the images used throughout the lab were provided from professor Wilson. Next by clicking on the raster tab and then right clicking on the image and selecting the inquire box a white square box will be displayed. The box was placed in the Eau Claire/ Chippewa area by dragging with the left mouse button. Next, the "subset and chip" and "create subset image" tools were used to create a subset of the image, and this was then saved under a specific name. The area that was captured is depicted as figure 1 below.
In section two of the lab the same image from the first step was used again. Next, the subset image that was created before was brought in an additional viewer and the file type was changed from (img.) to (shp.). After that the shapefile was overlaid on top of the eau_claire_2011.img. Next the shapefile was selected to show the area of interest by clicking on the two counties, Chippewa and Eau Claire one after the other, the areas should turn from blue to yellow. From here the area of interest is saved as an (aoi.) file and saved into the specific folder that was created at the beginning of the lab. By clicking on raster, then "subset and chip" as in step one the section is brought into the subset window. Figure 2 illustrates a screen capture of the finished product.
Part 2
The goal of part 2 is to create a higher spatial resolution image from one that is more course in order to optimize the viewers experience. The image ec_cpw_2000.img was brought into the viewer and a second viewer with ec_cpw_2000pan.img was also brought in. From here the pan sharpen tool was used to sharpen the image from 30 to 15 meters. This control is found under the raster tools, pan sharpen and on the pull down menu resolution merge. From here under the resampling techniques, nearest neighbor was selected. An image fusion folder was created for the output images. The output images are a layered photo of both the multispectral and panchromatic bands.
Part 3
Part 3 has the goal of using radiometric enhancements techniques in order to remove haze from an image. The haze reduction tool was used and a specific folder was created for the output image. For this exercise all the default values were used and a second viewer was opened in order to be able to see the differences between the images.
Part 4
The goal of part 4 was to use a recent development in Erdas that allows the user to synchronize Google Earth imagery from GeoEye (high resolution satellite) with the Erdas platform. By syncing the views it is clear to see the potential uses for this, as google earth is very good resolution.
Part 5
Part 5 involved resampling some images, which is the process of changing the size of the pixels. An image can either be resampled up or resampled down though there is no use to resample down. The image was brought in and fit to frame, then under the spatial tab in the top right on the pull down menu resample pixel size was selected. Again, a special folder was created to save the output images, the output cell size was changed from 30x30 to 15x15 meters and the default nearest neighbor method was accepted. Next, the same process was ran again, this time bilinear interpolation was selected rather than nearest neighbor and the bilinear interpolation can out much better.
Part 6
The goal of part 6 is to use image mosaicking to look at multiple different satellite images as one. Both the images were capture in May 1995 so the images were taken at roughly the same time. When bringing in both of the images, be sure to click on multiple images in virtual mosaic and make sure that background transparency is also checked. It is important to note that the images will be brought in one at a time but with the same steps as described above. The next task was to use Mosaic Express to create one seamless tile. The mosaic express button was selected and both of the images were added and saved into another specific new folder in the lab 4 folder. The default parameters were accepted and the model was then ran. The results are shown below.
The next step of this part focuses on using a much more advanced mosaic routine labeled MosaicPro in order to cut down on the amount of differences where the images are stitched together. Once the images were brought in, the order in which they were displayed was manipulated to see which would be better. In the color corrections tab the histogram matching was set to overlap areas. Then by clicking on the set overlap functions key the default was selected which was overlay. The mosaic was processed and the results were much better than the first way the mosaics were done.
Part 7
Part 7 focuses on binary change detection and image differencing. The change in brightness values from 1991 to 2011 for Eau Claire county and four surrounding counties is what is being looked at. The images were brought into Erdas in two separate viewers. Under the raster tab, two image functions was selected in order to get to two input operators, this is the tool that performs the operations on the images. The image differencing was saved in a new folder in the lab 4 folder. Be sure to change the operator from + to -. From there open up the metadata and the histogram to see the results.
Next a map of the changes from each image was made using spatial modeler. The equation highlighted below in orange shows how the negative values were removed from the differences in the images.
ΔBVijk = BVijk(1) – BVijk(2) + c
ΔBVijk(1)= Brightness values of 2011 image.
BVijk(2) = Brightness values of 1991 image.
Part 1
The first step to this lab is to open Erdas Imagine and then to set up a designated area to save the project that has a specific title including the users last name so it will not be mixed up. From there the image eau_claire_2011.img was brought into Erdas, it is important to note that all the images used throughout the lab were provided from professor Wilson. Next by clicking on the raster tab and then right clicking on the image and selecting the inquire box a white square box will be displayed. The box was placed in the Eau Claire/ Chippewa area by dragging with the left mouse button. Next, the "subset and chip" and "create subset image" tools were used to create a subset of the image, and this was then saved under a specific name. The area that was captured is depicted as figure 1 below.
 |
| Figure 1 |
In section two of the lab the same image from the first step was used again. Next, the subset image that was created before was brought in an additional viewer and the file type was changed from (img.) to (shp.). After that the shapefile was overlaid on top of the eau_claire_2011.img. Next the shapefile was selected to show the area of interest by clicking on the two counties, Chippewa and Eau Claire one after the other, the areas should turn from blue to yellow. From here the area of interest is saved as an (aoi.) file and saved into the specific folder that was created at the beginning of the lab. By clicking on raster, then "subset and chip" as in step one the section is brought into the subset window. Figure 2 illustrates a screen capture of the finished product.
 |
| Figure 2 |
Part 2
The goal of part 2 is to create a higher spatial resolution image from one that is more course in order to optimize the viewers experience. The image ec_cpw_2000.img was brought into the viewer and a second viewer with ec_cpw_2000pan.img was also brought in. From here the pan sharpen tool was used to sharpen the image from 30 to 15 meters. This control is found under the raster tools, pan sharpen and on the pull down menu resolution merge. From here under the resampling techniques, nearest neighbor was selected. An image fusion folder was created for the output images. The output images are a layered photo of both the multispectral and panchromatic bands.
Part 3
Part 3 has the goal of using radiometric enhancements techniques in order to remove haze from an image. The haze reduction tool was used and a specific folder was created for the output image. For this exercise all the default values were used and a second viewer was opened in order to be able to see the differences between the images.
Part 4
The goal of part 4 was to use a recent development in Erdas that allows the user to synchronize Google Earth imagery from GeoEye (high resolution satellite) with the Erdas platform. By syncing the views it is clear to see the potential uses for this, as google earth is very good resolution.
Part 5
Part 5 involved resampling some images, which is the process of changing the size of the pixels. An image can either be resampled up or resampled down though there is no use to resample down. The image was brought in and fit to frame, then under the spatial tab in the top right on the pull down menu resample pixel size was selected. Again, a special folder was created to save the output images, the output cell size was changed from 30x30 to 15x15 meters and the default nearest neighbor method was accepted. Next, the same process was ran again, this time bilinear interpolation was selected rather than nearest neighbor and the bilinear interpolation can out much better.
Part 6
The goal of part 6 is to use image mosaicking to look at multiple different satellite images as one. Both the images were capture in May 1995 so the images were taken at roughly the same time. When bringing in both of the images, be sure to click on multiple images in virtual mosaic and make sure that background transparency is also checked. It is important to note that the images will be brought in one at a time but with the same steps as described above. The next task was to use Mosaic Express to create one seamless tile. The mosaic express button was selected and both of the images were added and saved into another specific new folder in the lab 4 folder. The default parameters were accepted and the model was then ran. The results are shown below.
 |
| Figure 3 |
The next step of this part focuses on using a much more advanced mosaic routine labeled MosaicPro in order to cut down on the amount of differences where the images are stitched together. Once the images were brought in, the order in which they were displayed was manipulated to see which would be better. In the color corrections tab the histogram matching was set to overlap areas. Then by clicking on the set overlap functions key the default was selected which was overlay. The mosaic was processed and the results were much better than the first way the mosaics were done.
 | ||
Figure 4
|
Part 7 focuses on binary change detection and image differencing. The change in brightness values from 1991 to 2011 for Eau Claire county and four surrounding counties is what is being looked at. The images were brought into Erdas in two separate viewers. Under the raster tab, two image functions was selected in order to get to two input operators, this is the tool that performs the operations on the images. The image differencing was saved in a new folder in the lab 4 folder. Be sure to change the operator from + to -. From there open up the metadata and the histogram to see the results.
Next a map of the changes from each image was made using spatial modeler. The equation highlighted below in orange shows how the negative values were removed from the differences in the images.
ΔBVijk = BVijk(1) – BVijk(2) + c
ΔBVijk(1)= Brightness values of 2011 image.
BVijk(2) = Brightness values of 1991 image.
c = constant: 127 in this case.
I = line number
J = column number
K= a single band of Landsat TM
I = line number
J = column number
K= a single band of Landsat TM
 |
| Figure 6 |
Figure 6 above is a screenshot of the final image after running the image differencing. The red illustrates the changes from 1991 to 2011 and the grey had no change.
Results
Completing this lab led to increases in many of the skills that were introduced prior to completion of the lab. This lab covered a number of tools and processes that will have use in the long term. Using subset and chip, smaller images can be taken out of larger originals. Using the pan sharpen tool allows for the image to be more pleasing to the viewer and better for image interpretation, there is significantly less of a pixelated look. Additionally haze reduction is a tool that should be used almost any time there is aerial imagery because there is particulates and water vapor in the atmosphere that creates the haze that is seen. After haze reduction the image showed up darker and more defined, the white was gone and the bodies of water appeared much darker and clearer. Using google earth along with Erdas is also a critical tool. With the viewers linked if the Erdas image becomes too blurry, the google earth imagery is still very clear, this is essential in image interpretation. When looking at resampling, nearest neighbor did not show a large difference. In contrast the bilinear interpolation was a significant difference, the image was much sharper and clearer. The mosaic express tool was effective, though it was not ideal because there was still a clear line between the pictures. In comparison the MosaicPro had a smooth and unnoticeable transition between the images and the colors were also much closer as shown above in figure 4.
In conclusion was the image differencing from 1991 to 2011. The majority of the changes occurred not near urban center but more in rural areas. This could be because of a change in land use from agriculture to residential or vice versa, along with countless other possibilities.
Completing this lab led to increases in many of the skills that were introduced prior to completion of the lab. This lab covered a number of tools and processes that will have use in the long term. Using subset and chip, smaller images can be taken out of larger originals. Using the pan sharpen tool allows for the image to be more pleasing to the viewer and better for image interpretation, there is significantly less of a pixelated look. Additionally haze reduction is a tool that should be used almost any time there is aerial imagery because there is particulates and water vapor in the atmosphere that creates the haze that is seen. After haze reduction the image showed up darker and more defined, the white was gone and the bodies of water appeared much darker and clearer. Using google earth along with Erdas is also a critical tool. With the viewers linked if the Erdas image becomes too blurry, the google earth imagery is still very clear, this is essential in image interpretation. When looking at resampling, nearest neighbor did not show a large difference. In contrast the bilinear interpolation was a significant difference, the image was much sharper and clearer. The mosaic express tool was effective, though it was not ideal because there was still a clear line between the pictures. In comparison the MosaicPro had a smooth and unnoticeable transition between the images and the colors were also much closer as shown above in figure 4.
In conclusion was the image differencing from 1991 to 2011. The majority of the changes occurred not near urban center but more in rural areas. This could be because of a change in land use from agriculture to residential or vice versa, along with countless other possibilities.
Sources
Doctor Cyril Wilson, University of Wisconsin Eau-Claire, Spring 2017, Geography 338
