Sorry, you need to enable JavaScript to visit this website.

Part 2: Creating a Multispectral Mosaic Dataset

Part 2: Creating a Multispectral Mosaic Dataset

Quickbird is a high-resolution commercially-owned (DigitalGlobe) “earth-orbiting” satellite that is used to collect high-accuracy imagery; this imagery is available in both panchromatic (black and white) and multispectral (a wide range of frequencies across the electromagnetic spectrum). While the data used in this example is unavailable for testing, some sample Quickbird imagery can be downloaded from the FTP site at the Global Land Cover Facility by University of Maryland, College Park.

To work with this imagery we will be utilizing a mosaic dataset. This powerful way to store and manage raster data is extremely effective as it provides the user with added functionality to enhance the imagery and produce a polished, high quality final product.

First, create new mosaic dataset within a file geodatabase by right clicking on the file geodatabase in ArcCatalog, select New > Mosaic Dataset in the context menu. This will open the Create Mosaic Dataset tool where details about the input imagery will be specified. The first step is to name the mosaic dataset; the best practice for this would be to select a name that specifies the collection area, type of data and the date of collection (ie. GatewayNRA_Quickbird_Color_2012). Next, use the following directions to complete the remainder of the tool:

  1. Coordinate System = Choose based on Area of Interest (NAD 1983 UTM Zone 18N  was used for this example based on the area of interest - Gateway National Recreation Area)
  2. Product Definition = QUICKBIRD 4BANDS
  3. Click OK to run.

To add rasters to the mosaic dataset, right click on the mosaic dataset in ArcCatalog, and choose Add Rasters from the context window. Within this tool fill in the following specifications:

  1. Raster type = Quickbird
  2. Input Data = Workspace
    • Then click the folder icon and add the folder that contains the imagery.
  3. Check: Update boundary, Update overviews (this automatically checks Update Cell Size)
  4. Under Advanced Options:
    • Coordinate System = select same Coordinate system as previously (in this case NAD 1983 UTM Zone 18N)
    • Input Data Filter = *.IMD (green circle below)
    • Add New Datasets Only = Allow Duplicates
    • Check: Build raster pyramids, Calculate statistics
  5. Click OK to run the tool.

Next we need to Build Footprints. Footprints are calculated based on each raster image that was input into the mosaic dataset. Footprints help to determine which data is displayed in the mosaicked image. To do this, right click on the Mosaic dataset again, and choose Modify > Build Footprints; once the tool is open:

  1. Computation Method = Radiometry
  2. Minimum data Value = 1
  3. Maximum data Value = Default value (this changes based on input imagery)
  4. Approximate Number of Vertices = 30
  5. Check: Maintain edges, skip overviews, update boundary.
  6. Simplification Method = None
  7. Request Size = -1
  8. Minimum Region Size = 100
  9. Click OK to run the tool.

Now you’re ready to build seamlines. These help to ensure that the mosaicked image will appear to the highest quality, as they define the lines along which the images are “stitched” together. To do this, right click on the mosaic dataset again and choose Enhance > Generate Seamlines. Once open fill in the following options:

  1. Computation Method = Radiometry
  2. Sort Method = North_West
  3. Leave everything else as default
  4. Click OK to run the tool.

 

Now it is necessary to right click on mosaic dataset in ArcCatalog and select Properties from the context menu. Navigate to the Defaults Tab.

  1. Within the Image Properties section, set Allowed Mosaic Methods to seamline by unchecking everything else. Also leave seamline as the default method.
    • To do this, click on the  button.
  2. Set Default Mosaic Operator to Blend by clicking the down arrow.
  3. Click Apply, then OK.

Now we need to build overviews; right Click on the mosaic dataset and choose Optimize > Build Overviews. Leave all values as default and click OK to run.

The next step of the workflow is optional. We will now pansharpen the imagery. This is done by combining the multispectral (low resolution) and the panchromatic (high resolution) bands to create a high-resolution color mosaic dataset. Before you are able to complete this portion of the workflow, please be sure to create your panchromatic mosaic dataset using the Part 1: Creating a Panchromatic Mosaic Dataset blog post. For a more in-depth description of Pansharpening and functions, visit the Part 3: Create a Pansharpened Image Using ArcGIS blog post.

First, open Mosaic Dataset Properties by right clicking the dataset in ArcCatalog. Once the properties dialog box is open, go to the Functions tab.

  1. Right click on the Mosaic Function and insert the Pansharpening Function.
    • Under the Pan Sharpen tab in the dialog box, navigate to and select the panchromatic mosaic dataset.
    •  Leave the Multispectral mosaic dataset box filled with the default text of <Mosaic Function.OutputRaster>.
    • In the Infrared dataset box, delete the text, leaving the box blank.
    • Check the box to use the fourth band of the MS image as the infrared image.
    • Under properties: 
      • Method = Gram-Schmidt
      • Sensor = Quickbird
      • Leave the rest default.
    • Click OK to exit the function dialog box.
  2. Click Apply, then OK in the Mosaic Dataset properties to apply the new function to the mosaic dataset.

 

The final step to create a high-quality color mosaic dataset using Quickbird imagery is to Color Balance the mosaic dataset.

However, first it will be necessary to Build Pyramids and Statistics for all rasters that were used to create the mosaic dataset. To do this, right click the mosaic dataset in ArcCatalog, and under Modify select the Build Pyramids and Statistics tool.

  1. For the input data, navigate to one of the folder containing the individual raster tiles that make up the mosaic dataset.
    1. In the case of this example, you will need to run this tool twice due to the fact that the raster source images are located in two separate folders (refer to the Add Rasters section).
  2. Check Boxes: Include Sub-directories, Build Pyramids, Calculate Statistics, and Skip Existing.
  3. Leave the rest of the dialog as default values.
  4. Click OK to run.

Now, in the ArcCatalog, right click the mosaic dataset, and under Enhance select Color Balance.

  1. Select the correct mosaic dataset as the input.
  2. Balance Method = Dodging
  3. Color Surface Type = Color_Grid
  4. Leave the rest of the tool as defaults.
  5. Click OK to run.

Finally, you will need to re-build overviews. Right click the mosaic dataset and select Optimize > Build Overviews. Click OK to run with all default values.

 

This blog posting was developed with the support of a competitive grant (cooperative agreement number P09AC00212; task agreement number P13AC00875) from the National Park Service in partnership with the North Atlantic Coast Cooperative Ecosystems Studies Unit. It is part of a larger document available for download on the IRMA Portal.​

Main Topic

 
 

Authored by

erica
 

Last updated

09/10/2015 - 08:42