Lab 4: Miscellaneous image functions

Goal/ Background:

The overall objective of this lab was to familiarize ourselves with a few of the essential techniques needed in the remote sensing world. We would learn how to resample, mosaic images, generate a subset, link satellite image with Google Earth, and lastly expose us to the use of binary change detection. After this lab we now have a better understanding of the applications ERDAS can do and a better grasp on how to accomplish some of these tasks.

Methods:

Figure 1.1: creating a subset area for slipping out
an area of interest.

Figure 1.2: subset and clip tool options. Allows for selection
of subset area, AOI file, and type of radiometric resolution

Our first objective was to generate a subset of the Eau Claire area from the full Eau Claire regional image. We where able to accomplish this in two ways. The first way was to use an inquire box, which allows the user to determine their own square size by dragging the box over the area of interest. Once you have the desired box size around your AOI you then use the raster tools subset and set the subset based on the inquire box. However, since most of the time you will not be fortunate enough to have an AOI that is square we next needed to learn how to subset an image using a shape file. This is a more common method of sub setting an image for it allows you to get an exact design of an area. First we needed to bring in the shape file over the image we where trying to subset. Then we selected the shape file area and saved it as an AOI file. Once you have your new AOI you then can use the raster tool subset and chip (create subset image) but this time instead of applying an inquire box you will use the AOI feature on the bottom of the tool and select the recently saved AOI this will then generate a subset based on the AOI you created.

Objective two:

The second task in our lab was to learn a very common technique known as pan sharpening. to accomplish this we will be using images of the Eau Claire area taking in 2000. the first image is our multi spectral image with a spatial resolution of 30x30 meters. and the second image that we will use for pan sharpening is of the same area but has a spatial resolution of 15x15 meters. In order to complete this task we needed to merge the two images resolutions together by going into raster tools pan sharpen and resolution merge. Within the resolution merge tool you have a few steps that you need to determine.

1) you need to decided the two images you want to merge.

2) you can pick what type of method for merge you can do such as principal component, nearest neighbor, brovey transfer
3) you thirdly chose what type of resampling technique such as nearest neighbor, bi-linear interpolation, cubic convolution.
Once you have chosen these parameters you now can run your merge and the outcome will generate a new higher resolution image.

Objective three:

Haze reduction is one of the few key tools that every person who within the remote sensing field should know how to do. It is a process that uses algorithms to cut down on the amount of atmospheric scattering. To achieve this goal you will need to open up raster tools under radiometric tab use the haze reduction tool. The tool is very simple you simple need to pick the image you want to be run through the algorithm and let the computer process out all areas that meet the criteria. The overall outcome will be a sharper contrasted image with areas that once where full of haze now sharpened and clear. The outcome can be contrasted when looking at the differences in figure 1.3 the original photo and figure 1.4 the corrected haze reduction photo.

Figure 1.3: Map of eau Claire area prior to haze reduction enforced.
not high reflection in lower right corner

Figure 1.4: corrected image after haze reduction

Objective four: 

The fourth task of this lab is a cutting edge tool that has only recently been introduced into remote sensing world which allows you to synchronize your image with google earth and allow you to have a selective key when interpreting data. To sync the images you simple open your image into a viewer. Then by opening the google earth function and using the match GE to view mode it will automatically zoom into the extent of the image you have pulled up. From there as you zoom in and out the two images will be synced and google earth with adjust accordingly. The resulting sync is very usefully for any application in which you are looking for a very spatially high resolution selective key to help analyze any data. The figure below shows the ability of Google earth to connect with ones image and allow for a high resolution selective key application.

Figure 1.5: shows ERDAS new ability to sync with Google earth giving a new dimension to selective high resolution keys.

Objective five: 

Part five of lab four introduced a new topic to our class of re-sampling. In the last week of class we recently have gone over the types of both re-sampling up making pixel size smaller and re-sampling down making pixel size larger. Along with the four different methods that can be used to re-sample the pixels. For this lab we will be using both nearest neighbor and bi-linear interpolation and re-sampling up both times. 

For both methods the task of re-sampling is exactly the same other than selecting the method type within the tool set. You first open raster tools and under spatial tools select resample by pixel size. 

From there you then can input any image you wish to resample by in putting it into the input file, saveing it to the desierd folder with name of your choosing. select what type of resample method you desire and change your pixel size to what ever size you want below the methods. One thing to keep in mind is to select the square cells options to make sure the pixels will come out as uniform squares in the re-sample. Then run the program and admire the outcome. In or project you could only notice a slight difference in clarity of the bilinear interpolation versus the nearest neighbor but it was not any drastic change from the origianl for we only cut the spatial resolution in half. 

Image of resampling

Objective six:

Part six of this lab was to introduce us to the different methods of mosaicking in remote sensing. Mosaicking is a very useful tool for many people in the remote sensing world for a lot of the time you will be dealing with an area of interest that may not necessarily lie within one image or may be such a large area that it requires more than one photo to capture the entire area. To successfully mosaic one must first make sure all the images they will combine have similar temporal and spectral resolution or the images will not match up well. then they need to decided what method they will use. If they want a quick but a little more messy mosaic express will be a good option for it is a very computationally fast application but the results are not aesthetically pleasing. the second option and the more widely used is mosaic pro. Which allows for a little more fine detail selection. It will allow the operator to select what image he or she wants on top (typically the higher resolution) and also allows for one to select a better histogram match up which will allow for a high clarity and better match up of the images. figure 1.6.

Figure 1.6: mosaic pro menu options. Fx ( allows operator to adjust seam line function and overlap options)
histogram tool allows you to set histogram matching making image more clear and color tone clarity enhanced.

Objective 7

The last objective in this lab was to produce a binary change map that showed all areas within the Eau Claire area that experienced urbanization from 1991 to 2011. To do this we needed to run a binary change algorithm of the two images to isolate the areas in which experienced change. To do this we first needed to find all areas that where on the outer half of the standard deviation curve which in this case was any point that lied above 208 on the radiometric scale. Then by using a model we could highlight all of these areas using an either or expression. It read EITHER 1 IF (image from 1991 > 208 ( which was the change/no change threshold) OR 0 OTHERWISE. The express basically will isolate and highlight an area that lies above the threshold in this case was all areas that experienced urbanization and apply a 1 value which will show up on the map and a 0 or no value making all other areas blank. The end result is an image that only shows urbanization over time.

Once we had our final image we then could import that information into arc map and generate a map showing the results figure 1.7

Results:

Figure 1.7: shows all regions in Eau Claire area that have shown urbanization from 1991 to 2011 in red

The overall outcome of this lab was to introduce us to a variety of essential tools within the remote sensing realm. Tools like sub setting, haze reduction, binary change amongst others will all be key to the advancement of our remote sensing careers, and because of this lab we now have the ability to carry out such applications. I can now say that I have a better understanding of how some of these tools work and in what situations certain tools should be applied. I enjoyed this lab and hope that the theme of applied knowledge will continue throughout the weeks to come.

Sources:

University of Wisconsin Eau Claire Geography Department