Part 2—Process and Analyze the Images
Step 1 Launch ImageJ and Open the First Image
- Launch ImageJ by selecting it from the Start Menu or by double-clicking the icon that looks like a microscope. Once ImageJ is running, the following menu and tool bar will be displayed.
- To load the first image, choose File > Open, navigate to the Forests folder, and select image_1994.tif.
- This is what the image looks like when it opens. The roads, the dark bands, are cutting across the forest like the tributaries of a river.
Step 2 Scale and Process the Image
In these satellite images, each pixel (picture element, the smallest unit in an image) covers a square of the Earth's surface that is just 12.5 m x 12.5 m in size. That is a lot of detailed information in one very small space. In order to make the image easier for the computer to handle, you will use ImageJ to change the resolution to 50 m x 50 m so that each pixel in the image covers more surface area.
- Select Image > Scale... from the menu bar and the Scale window opens. To reduce the scale by 25%, input .25 for the X Scale and .25 for the Y Scale. Select Bilinear Interpolation (default) and check Create new window. Title the new image map1.tif and click OK.
- Select Image > Scale... from the menu bar.
- The Scale window opens. To reduce the scale by 25%, input .25 for the X Scale and .25 for the Y Scale. Select Bilinear Interpolation (default), and check Create new window. Title the new image map1.tif. Then click OK.
- The resulting image is now a lot smaller than the original one. Additionally, it looks a little smoother.
- Select Analyze > Set Scale.... Notice how .250 is showing as the Distance in Pixels. This is because the image was reduced by 25%. To obtain accurate measurements, the scale must be reset. Change the Distance in Pixels to 1.0 and click OK.
- Close the original image (image_1994.tif).
- Extract the deforested areas out of the image by applying the FFT Bandpass filters. Filter Large Structures Down to 40 pixels and Filter Small Structures up to 3 pixels. Check Autoscale After Filtering and Saturate Image When Autoscaling. Then click OK.
- Select Process > FFT > Bandpass Filter...
- An FFT Bandpass Filter window opens. Filter Large Structures Down to 40 pixels and Filter Small Structures up to 3 pixels. Check Autoscale After Filtering and Saturate Image When Autoscaling. Click OK.
- The bandpass filter makes the image even smoother, so that the image does not look noisy anymore. Your bandpass filtered image should look like this.
Step 3 Threshold and Measure the Image
- Divide the forest image into two classes using thresholding. Select Image > Adjust > Threshold. The Threshold window opens. Segment the image using the Maximum Entropy thresholding method. Click Apply.
- Measure the area of the image that is deforested. First, make sure that Area and Area Fraction are chosen in the Set Measurements window. Then choose Analyze > Measure. The Results window displays the measurements. In this case, it shows the total area of roads (in pixels) and the percent of the total area that is occupied by roads. Write these measurements down on a piece of paper. Do not close the Results window.
- Choose Analyze > Set Measurements.
- The Set Measurements window opens. Check Area and Area Fraction to collect those measurements. Click OK. Note: When you work with the 1995 and 1996 image, you will not need to set these measurements again. ImageJ will continue to collect these measurements until you choose other options.
- Choose Analyze > Measure.
- The Results window shows the total area of the image (in pixels) and the percent of the total area that is occupied by roads. Write these measurements down on a piece of paper. Do not close the Results window.
The threshold is a statistical value used to divide the image in two classes. In this case, these classes approximately correspond to deforested and forested areas.
Note: It is possible to select File > Save As... to export your results to an Excel file. However, since you only have two measurements, this is not necessary. Also, if you keep the Results window open, then additional measurements will accrue there.
Step 4 Save the Processed Image
- Make map1.tif active by clicking once on it. Then choose File > Save As.., navigate to your Forests folder and save the image.
- Close map1.tif.
Step 5 Prepare Additional Images
- Repeat Steps 1 to 4 above for the 1995 and 1996 images.
- At the end you should have three images labeled with their file names map1.tif, map2.tif, map3.tif .
- You should also have three measurements indicating the percent of the area that was deforested in 1994, 1995, and 1996.
Step 6 Compare Deforestation Across All three Years
- What was the percent deforestation in 1994? in 1995? in 1996?
- How has the fraction changed through the years?
It increased from 15.50% to 23.12%.