Image Change Detection widget

The Image Change Detection widget computes the difference between two input imagery layers and generates a new imagery layer containing the change information.

Change detection is one of the fundamental applications in imagery and remote sensing. It is the comparison of multiple imagery layers, typically collected for one area at different times, to determine the type, magnitude, and location of change. Change can occur because of anthropogenic activity, abrupt natural disturbances, or long-term climatological or environmental trends.

The output from change detection is a difference raster where each pixel contains the type or magnitude of change.

Examples

Use this widget to support app design requirements such as the following:

• You want to understand changes in vegetation density and plant health.
• You want to monitor changes related to water content in water bodies by comparing remote sensing images taken before and after a flood.
• You want to study water temperature changes by comparing imagery layers containing data recorded at different ocean depths.

Usage notes

This widget requires connection to a Map widget. The Map widget must contain a web map or web scene with dynamic imagery layers.

Note:

The Image Change Detection widget only supports dynamic imagery layers.

When you include this widget in an app, the widget panel provides users with the following interaction options:

• Under Input imagery, you can select input layers.
  • Use the drop-down menus to select a From imagery layer and a To imagery layer from the map or scene.
  • In the widget's settings, you can preset input imagery layers. If you do, then at run time users can only view the present layer names, not change them.
• Under Change detection settings, you can choose a change detection method and set its parameters. The Image Change Detection widget supports three change detection methods:
  • Spectral change methods—Compute change based on spectral angle or distance for two multi-band imagery layers. The spectra of each pixel is treated as a vector.
    • Spectral Euclidean Distance—Measure the Euclidean distance between the two vectors.
    • Spectral Angle Difference—Measure the angle (in radians) between the two vectors.
  • Image index change methods—Calculate a band index for two multi-band imagery layers, then compute pixel value change.
    • Vegetation—Use Normalized Difference Vegetation Index (NDVI) as the indicator. This requires red and near-infrared bands.
    • Water—Use Normalized Difference Water Index (NDWI) as the indicator. This requires green and near-infrared bands.
    • Burn scar— Use Normalized Burn Ratio Index (NBR) as the indicator. This requires near-infrared and shortwave infrared bands.
    • Build-up—Uses Normalized Difference Build-up Index (NDBI) as the indicator. This requires near-infrared and shortwave infrared bands.
  • Pixel value change methods—Compute absolute or relative change of a selected band between two imagery layers.
    • Absolute difference type—Measure the mathematical difference between pixel values.
    • Relative difference type—Measure the difference between pixel values calculated in relation to the magnitude of the maximum pixel values being compared.
• Under Result layer, you can enter a name for the result layer. This option is available when you choose Multiple layers under Result layer mode in the widget's settings.
• Click Run to run the analysis, create a new result layer, and add it to the map.
• After you run the analysis, the Layer display settings section appears. This section includes the following additional interaction options:
  • Result layer—The result layer or layers appear here for analyses run with image index change methods. Multiple result layers can appear here if you choose Multiple layers under Result layer mode in the widget's settings. Click a result layer to view it on the map.
  • Show difference within—Use this slider to add a pixel value masks that highlights the changes. This setting appears for result layers generated from image index change methods.

  Tip:

  You can use a Map Layers widget to change the order and visibility of layers generated at run time.

Settings

The Image Change Detection widget includes the following settings:

• Select a Map widget—Select a Map widget.
• Preset imagery layers—Optionally, set default input imagery layers. If you do, these layers are unchangeable by the user at run time. If you do not, users can select input imagery layers at run time.
  • From imagery layer—Preset a default From imagery layer.
  • To imagery layer—Preset a default To imagery layer.

  Note:

  You can use the Image Collection Explorer widget to view all of the individual images in an image collection from a dynamic imagery layer and create new imagery layers to use as From and To layers with the Image Change Detection widget.

• Result layer mode—Choose a result layer mode:
  • Single layer—There can only be one result layer at a time. Re-running the analysis replaces the existing result layer.
  • Multiple layers—Allow for there to be multiple result layers if the user runs the analysis multiple times. Re-running the analysis adds the new result layer to the map.
• Change detection method—Choose which image change detection methods to make available to users at run time. Click the Configure image index button to choose which indices to make available at run time.
  • If you preset a From imagery layer, To imagery layer, or both, drop-down menus appear next to each index in the Configure image index panel. You must select the corresponding bands for each preset input layer.