Lab 4: Miscellaneous image functions

Goal and Background:

The purpose of this lab was to use the various image functions learned in class in the ERDAS imagine software. By doing so, we were able to gain skills in image processing, image enhancing, selecting an area of interest, and creating mosaics with multiple methods. The satellite images used in his lab was provided by Earth Resources Observation and Science Center, United States Geological Survey. Shapefile is from Mastering ArcGIS 6th edition Dataset by Maribeth Price, McGraw Hill. 2014.

 

 

 

Methods:

Part 1: Image subsetting (creation of area of interest AOI) of study area

            To create an area of interest in ERDAS Imagine I used an inquire box. By putting the inquire box over the study area, which in this case was the Eau Claire- Chippewa area, and then creating a subset image of the area I was able to create a new file of AOI. The create subset image tool can be found in the Raster tab under Subset & Chip. This can be seen in Figure 1.  

Another method used to create a subset of an area of interest is through the use of a shapefile. This is often used due to the fact that AOIs are usually not a perfect square or rectangle. In this case, I used the shapefile of the Chippewa and Eau Claire counties. By uploading both the shapefile and image at hand I was able to create a AOI layer and create a subset image. This can be seen in Figure 2.



Figure 1: Inquire Box subset image

             

 

 

Part 2: Image fusion

            This section of the lab focused on the enhancing the spatial resolution of an image through pansharpening for interpretation purposes. To pansharpen an image, I took the initial multispectral image and a higher resolution image and conducted a resolution merge. The method used was multiplicative which is a programmed equation used by the software to conduct the merge. The resampling technique was set to nearest neighbor. This creates an overall higher resolution image with more detail to the features.



Figure 2: Shapefile subset image

 

Part 3: Simple radiometric enhancement techniques

            In this section, an enhancement technique was performed to increase the quality of aerial images. The technique was haze reduction. This tool can be found under the Raster tab and the radiometric tab. The haze reduction, as its name implies, created a clearer image.

 

Part 4: Linking image viewer to Google Earth

            This portion of the lab was dedicating to creating a link between Google Earth and a referenced aerial image. By linking the views, you are able to use Google Earth as a Selective Key during image interpretation. It allows you to find specific features of normal earth colors on Google Earth and in turn find the same feature on the image in ERDAS.

 

Part 5: Resampling

            This part of the lab focused on resampling images. This is the process of changing the size of the pixels in an image. Resampling up or resampling down can aid in the analytical process of the image. To begin the resampling process, I took the image of Eau Claire and under Raster> Spatial chose Resample Pixel Size. In the window, I filled in the parameters needed and set the technique to nearest neighbor and set the new pixel size to 15x15 m. The final product did not look much different from the original.

            The second method I used was Bilinear Interpolation. I followed the same process by setting the pixel size to 15x15m. However, the outcome of this method was much clearer when zoomed in and allowed for more detail.

 

Part 6: Image Mosaicking

            Creating an image mosaic can help when your AOI is on two separate images. Mosaicking takes two images and creates a singular image that is blended (meaning you cannot distinguish where one image ends and the other begins easily). There are several programs in ERDAS that perform this function. Mosaic Express takes minimal user input. I only had to select the images I wanted to be mosaicked and changed the program to multiple. The result (figure 3) is not very good quality because of the obvious color differences. MosaicPro is another program in ERDAS. This program used histogram matching which allows for color correction. This is the reason MosaicPro has a more blended output (figure 4).

Figure 3: Mosiac Express Image

 

 

 

 

 

Figure 4: MosaicPro Image

 

 

Part 7: Binary change detection (image differencing)

            In this section of the lab, we were set with the task to determine areas of change from 1991 to 2011 in the Chippewa Valley. To do so, I first had to conduct binary change detection. His is a process found in the Raster Tab under functions and two image functions. After running the image differencing, I then checked the histogram and calculated the change/no-change value from he given statistics in the images metadata. This is calculated with mean + 1.5 standard deviation. The values can be seen in the histogram below (figure 5). I then used the Model Maker to run various calculations on images of the AOI. This is done to create a map showing the changed areas from 1991 to 2011. This map can be seen in figure 6.

Figure 5: Histogram with change/no-change values 

 

Figure 6: Areas of change

Results:
          The results of the various tools and processes can be seen in figures 1-6. Image 1 shows the subset image gotten from the inquire box. Image 2 shows the subset image gained from clipping it with a shapefile. Image 3 shows the mosaicked image through Mosaic Express whereas image 4 shows the mosaicked image created through mosaic express. Image 5 shows the histogram used to determine the area of change and no change. The final result of the binary change detection can be seen in figure 6. As seen, the areas of change are largely found in agricultural areas.    

 

Sources:

Earth Resources Observation and Science Center, United States Geological Survey

Mastering ArcGIS 6th edition Dataset by Maribeth Price, McGraw Hill. 2014