envi - image change detection

Image Change TutorialIn this tutorial, you will use the Image Change workflow to compare two images of anarea over Indonesia that was impacted by the December 26, 2004 tsunami. The firstimage is a before image, taken in April, 2004. The second image was taken inJanuary, 2005.

The first image shown below is from tsunami_before.dat, and the second imageshows tsunami_after.dat in a Standard Portal. You can see in the portal that thereare substantial differences between the two images when you adjust theTransparency sliders on the toolbar. Specifically, many vegetation areas werewashed out by the tsunami.

of 12© 2015 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation. All Rights Reserved. Thisinformation is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export

Administration Regulations (EAR). However, this information may be restricted from transfer to various embargoedcountries under U.S. laws and regulations.

Image Change Tutorial

References

Image Change Detection:

Normalized Difference Vegetation Index (NDVI): Jensen, J. R., 1986.Introductory Digital Image Processing, Prentice-Hall, New Jersey, p. 379.

Normalized Difference Water Index (NDWI): McFeeters, S.K., 1996. The use ofnormalized difference water index (NDWI) in the delineation of open waterfeatures, International Journal of Remote Sensing, 17(7):1425–1432.




Normalized Difference Built-up Index (NDBI): Zha, Y., J. Gao, and S. Ni, 2003.Use of normalized difference built-up index in automatically mapping urbanareas from TM imagery, International Journal of Remote Sensing, 24(3):583–594.

Burn Index: Burn Index uses an opposite Normalized Burn Ratio (NBR), which is-NBR. The NBR reference is Key, C.H.; Z. Zhu; D. Ohlen; S. Howard; R.McKinley; and N. Benson, 2002. The normalized burn ratio and relationships toburn severity: ecology, remote sensing and implementation. In J.D. Greer, ed.Rapid Delivery of Remote Sensing Products. Proceedings of the Ninth ForestService Remote Sensing Applications Conference, San Diego, CA 8-12 April,2002. American Society for Photogrammetry and Remote Sensing, Bethesda,MD.

Auto-thresholding:

Otsu's: Otsu, N. A threshold selection method from gray-level histograms. IEEETrans. Systems Man Cybernet. Vol. 9, pp. 62–66, 1979.

Tsai's: Tsai, W. Moment-preserving thresholding. Comput. Vision GraphicsImage Process. Vol. 29, pp. 377–393, 1985.

Kapur's: Kapur, J., Sahoo, P., Wong, A. A new method for graylevel picturethresholding using the entropy of the histogram. Comput. Vision GraphicsImage Process. Vol. 29 (3), pp. 273–285.

Kittler's: Kittler, J., Illingworth, J., Minimum error thresholding, Pattern Recogn.Vol. 19, pp. 41–47, 1986.

Files Used in this TutorialTutorial files are available from the Exelis VIS website or on the ENVI® Resource DVDin the change_detection directory.

File Descriptiontsunami_before.dat QuickBird image over Indonesia, April, 2004tsunami_before.hdr Header file for abovetsunami_after.dat QuickBird image over Indonesia, January, 2005tsunami_after.hdr Header file for above




http://www.exelisvis.com/Learn/Resources/Tutorials.aspx

Selecting Files for Image ChangeIn the File Selection panel, you choose the two images to include in image changedetection.

1. Start ENVI.

2. From the Toolbox, select Change Detection > Image Change Workflow.The Select File panel appears.

3. Click Browse next to the Time 1 File field. The Select Time 1 Input File dialogappears.

4. Click Open File. The Open dialog appears.

5. Navigate to change_detection , select tsunami_before.dat, and click Open.

6. Click OK.

7. Click Browse next to the Time 2 File field. The Select Time 2 Input File dialogappears.

8. Click Open File. The Open dialog appears.

9. Navigate to change_detection, select tsunami_after.dat, and click Open.

10. Click Next. The Image Registration panel displays.

11. Since the Time 1 and Time 2 images are co-registered already, keep the defaultselection of Skip Image Registration.

12. Click Next. The Change Method Choice panel appears.

13. Keep the default selection of Image Difference, and click Next. The ImageDifference panel appears.

Image Difference SettingsIn the Image Difference panel, set the parameters to use for the difference analysis.In this step, you perform an image difference analysis based on a band or featureindex. The feature index provides options to detect changes of a specific feature,such as vegetation, water, built-up areas, or fire burn areas. For the QuickBird dataused in this exercise, Vegetation Index (NDVI) and Water Index (NDWI) areavailable. Built-up Index and Burn Index are available only if if an image has ashortwave infrared band, such as Landsat data,




1. The Difference of Input Band option and Band 1 were selected by default.

2. In the toolbar Go To field, enter 746319.499,585303.471 and press the Enterkey on the keyboard. The view centers over the area.

3. Enable the Preview check box. A Preview Window appears. In the PreviewWindow, areas that decreased in the data value of the selected band appear asred, and areas that increased appear as blue.

4. With the Preview Window still open, enable the Difference of Feature Indexand keep Vegetation Index (NDVI) as the selected feature index.




5. Click Next. The difference analysis begins.

6. When image difference processing is complete, the difference image appears inthe Image window and the Thresholding or Export panel appears.

7. Keep the default selection of Apply Thresholding. This option allows you toset parameters that help the algorithm determine which areas have big change.When you select this option, you can export multiple outputs at the end of theworkflow. (If you select Export Image Difference Only, you will not be ableto select additional processing parameters, and you can only export thedifference image.)

8. Click Next. The Change Thresholding panel appears.




Change ThresholdingIn the Change Thresholding step, specify change you want to show between the twoimages. You can use pre-set auto-thresholding techniques, and you can manuallyadjust thresholding.

1. In the Auto-Thresholding tab, keep the default selection of Increase andDecrease. This option shows areas of increase (in blue) and decrease (in red).(If you are only interested in areas of vegetation decreased by the tsunami,select Decrease Only.)

2. In the Select Auto-Thresholding Method drop-down list, try selecting eachoption, one at a time, then examine the result in the Preview Window. The auto-thresholding choices are:

n Otsu's: A histogram shape-based method. It is based on discriminateanalysis and uses the zeroth- and the first-order cumulative moments ofthe histogram for calculating the value of the thresholding level.

n Tsai's: A moment-based method. It determines the threshold so that thefirst three moments of the input image are preserved in the output image.

n Kapur's: An entropy-based method. It considers the thresholding imageas two classes of events, with each class characterized by a ProbabilityDensity Function (PDF). The method then maximizes the sum of theentropy of the two PDFs to converge on a single threshold value.

n Kittler's: A histogram shape-based method. It works on approximatingthe histogram as a bimodal Gaussian distribution and finds a cutoff point.The cost function is based on the Bayes classification rule.

The References at the beginning of this tutorial provide additional informationabout the auto-thresholding methods.

3. In this exercise, we will use the default Otsu's thresholding method. Below isan example of the Preview Window with the Otsu's method selected.




4. You can also experiment with manually adjusting the threshold settings. To dothis, select the Manual tab.

5. Use the slider bars to adjust the Increase Threshold and DecreaseThreshold settings, then view the changes in the Preview Window.

6. When you are done experimenting with manual adjustments, click the Resetbutton to return to the default settings.

7. Click Next. When you click Next, the difference image will be classified into BigIncrease, Big Decrease and Other, based on the threshold values. TheCleanup panel appears.




Cleaning Up Image Change ResultsThe cleanup step refines the result. You can preview what the refinement will likelook before you apply the settings.

1. Keep the default selections for both cleanup methods:

n Enable Smoothing removes speckling noise.

n Enable Aggregation removes small regions.

2. Enter values for the cleanup methods:

n Specify the Smooth Kernel Size using an odd number (e.g., 3 = 3x3pixels). The square kernel's center pixel will be replaced with the majorityclass value of the kernel. Keep the value at 5.

n Specify the Aggregate Minimum Size in pixels. Regions with a size ofthis value or smaller are aggregated to an adjacent, larger region. Keepthe value at 100.

3. Preview the cleanup result before processing.




You can change the cleanup settings and preview the results again, if desired.

4. Click Next. The Export panel appears.

Exporting Image Change ResultsIn the final step of the workflow, you will save the output from the analysis.

To export results:

1. In the Export Files tab, enable the check boxes for the exports:

n Export Change Class Image saves the thresholding result to a rasterfile.




n Export Change Class Vectors saves the vectors created duringthresholding to a shapefile.

2. Use the default paths and filenames.

3. In the Additional Export tab, enable the check boxes for the remainingexports:

n Export Change Class Statistics saves statistics on the thresholdingimage.

n Export Difference Image saves the difference image to a raster file.

4. Use the default paths and filenames.

5. Click Finish. ENVI creates the output, opens the layers in the Image window,and saves the files to the directory you specified.

6. Select File > Exit to exit ENVI.

Copyright Notice:

ENVI is a registered trademark of Exelis Inc., a subsidiary of Harris Corporation.




envi - image change detection

Documents