Miscellaneous Image Functions

Goal

Background: The goal of this assignment was to familiarize the user with image delineation, spatial resolution optimization, radiometric enhancement, linking to Google Earth through Erdas Imagine 2013, and resampling.

Methods

Subsetting:
Purpose: Subsetting allows the analyst to clip a portion of an image that they want to focus on, rather than having to view the entire image.

Method 1: This portion of the assignment was concerned with image subsetting of a study area. The first method of subsetting is by creating a rectangular Inquire Box. Once an image has been added to the viewer, in this case an image of the Eau Claire-Twin Cities area, pick the Raster option from the top menu bar to activate the raster tools. Then, right-clicking on the image will let the analyst fit the image to the screen and then select the "Inquire Box". The Inquire Box was then enlarged and moved to focus on Chippewa and Eau Claire counties (Figure 1).

Figure 1: This picture was extracted from the assignment handout and
illustrates the area that was focused on, denoted by the white box.

Following this, the tool "Subset & Chip," then "Create Subset Image" were selected. Then the analyst selects an output for where the image subset will be sent to, clicks "From Inquire Box" so that the image that was focused on in the prior step will be extracted, and then the extraction is run. What is left is shown in Figure 2.

Figure 2: This image is what is output when the Image Subset
extraction is run using the "From Inquire Box" option.

Method 2: For this method, a shapefile of Eau Claire and Chippewa Counties was added to the same image used in the previous method. This shapefile shows up in the viewer as the outline of Eau Claire and Chippewa Counties. Next, both counties are selected together which is confirmed when the shapefiles turn from a blue to yellow color. These two sections make up the area of interest, or AOI, for this portion of the assignment (Figure 3). 

Figure 3: This picture was extracted from the assignment handout and
illustrates the shapefile that focused on, denoted by the yellow boxes.

After these boxes are selected, they are saved as an AOI Layer. Subset and Chip is selected under Raster again to open up the subset window. This time, however, after the output folder and file is selected, the option "AOI" at the bottom of the pop-up window is chosen. The function is run and the end result is shown in Figure 4. With this AOI subset, the image that is extracted retains the same shape as the AOI as selected by the analyst, while when using the first method a rectangular shape is the only option.

Figure 4: This image is what is output when the Image Subset
extraction is run using the "Area of Interest (AOI)" option. 

Image Fusion:

Purpose: Image fusion allows the analyst the ability to create a higher spatial resolution from a coarse resolution image, in order to decipher visual clues in a more precise manner.

Method: For this method, an image of the Eau Claire/Chippewa County area was used, as well as a panchromatic image of the Eau Claire/Chippewa County Area. The Raster tab was selected, followed by Pan Sharpen, and then Resolution Merge from the drop down menu. Once the Resolution Merge window was opened, the proper input and output files were designated and a multiplicative algorithm was assigned to properly denote how the image was going to be sharpened. The Resampling Technique selected was "Nearest Neighbor." After this, the function was run and the new image was produced. In order to see the difference between the original and the pansharpened image, the two images were linked and zoomed in, as shown in Figure 5. This process fused a 30 meter resolution image with a 15 meter resolution image, the result is an image that is slightly darker and crisper. The finer image darkens the outline of the preexisting image and makes the image much cleaner looking.

Figure 5: The image on the left is the original image.
The image on the right is the pansharpened, fused image.

Radiometric Enhancement Techniques:

Purpose: Radiometric Enhancement Techniques allow the analyst the ability to enhance image spectral and radiometric quality. The main purpose of this portion of the exercise was to learn how to reduce the presence of haze in an image.

Method: For this method, an image of the Eau Claire-Chippewa County area was loaded to the viewer screen. The Raster Tab was selected, followed by Radiometric, and then Haze Reduction from the drop down menu. An output location was designated on the Haze Reduction window that came up and the function was run. This method darkened the image and made the colors more vibrant in order to reduce the visual presence of haze. If enough attention is used, the haze is still there, though the deeper and more vibrant color pattern makes it much less noticeable than in the original image (Figure 6).

Figure 6: The image on the left is the original image with a prominent presence of haze.
The image on the left is the image that has undergone a haze reduction technique.

Linking Erdas Imagine 2013 to Google Earth:

Purpose: Linking Erdas to Google Earth allows the analyst the ability to reference a, potentially, much clearer image to use for referencing.

Method: For this method, an image of the Eau Claire-Chippewa County area was loaded to the viewer screen. The Google Earth Tab was selected, follow by "Connect to Google Earth." Then the "Match GE to View" option was selected. When zoomed in close enough, the pixels of the Erdas image are very noticeable and the image is not that easy to perceive. The Google Earth image, however, uses GeoEye and is much clearer, at much closer ranges, than the Erdas image (Figure 7).

Figure 7: This image was extracted from the assignment handout and shows the synced
viewer from Erdas Imagine 2013 on the left with the synced Google Earth viewer on the right.

Resampling:

Purpose: Resampling allows the analyst the ability to alter the pixel size of an image, in order increase or decrease the resolution of a given image.

Method 1: For this method, an image of the Eau Claire-Chippewa County area was loaded to the viewer screen. The Raster tab was selected, followed by Spatial, and Resample Pixel Size from the drop down menu. This opens the Resample window. After correctly filling out the input and output locations, select "Nearest Neighbor" under Resample Method, and change the output cell size from 30 meters to 20 meters. Following this, the function is run and the image is created (Figure 8). There is a noticeable difference in clarity between the two. The image on the left, the original image, is much less detailed than the one on the right. One area the analyst used to reference this was the islands in the center of Dells Pond. The island close enough to the downstream portion of the lake appears to not have much of a jagged perimeter. However, when the image is resampled and the pixel size is decreased there is a noticeable change in the jaggedness of the same island. 

Figure 8: This image shows the difference in detail when using the Nearest Neighbor
method, between the original image, on the left, and the resampled image, on the right.

Method 2: For this method, an image of the Eau Claire-Chippewa County area was loaded to the viewer screen. The Raster tab was selected, followed by Spatial, and Resample Pixel Size from the drop down menu. This opens the Resample window. After correctly filling out the input and output locations, select "Bilinear Interpolation" under Resample Method, and change the pixel size from 20 meters to 30 meters. Following this, the function is run and the image is created (Figure 9). This provided relatively the same amount of detail as did resampling by using the nearest neighbor technique. A noticeable difference being that the features seem much more streamlined and smooth when using the bilinear interpolation method.

Figure 9: This image shows the difference in detail when using the Bilinear Interpolation
method, between the original image, on the left, and the resampled image, on the right.

Results:

This assignment was very useful to begin to understand the various methods of troubleshooting and fixing the images that remote sensing analysts are provided with. Through this assignment, different tools were introduced and methods were conveyed that will prove very useful in the future when working with remote imagery that may be subpar or not up to industry standards.