Recently, I acquired two scenes of GeoEye imagery. They were provided as 7 different images. I opened the properties for each image, by clicking the raster product icon in the Catalog window and looked at the Key Metadata tab to know which images were collected together. I then organized each of the images into two folders named North and South. (This helps me remember which images belong together.)
Next, I created a new mosaic dataset. I defined a name and coordinate system, but also the GeoEye-1_4Bands Product Definition. I picked a Product Definition, because when I add the mosaic dataset to ArcMap, it will display the band names according to their wavelength; which I find very useful. Also, since I want to color balance this collection, it needs to be 8-bit, so I set the Pixel Type to 8-bit unsigned.
Then I added my images in two steps. I right-clicked the mosaic dataset in the Catalog window, and clicked Add Rasters to open the Add Rasters to Mosaic Dataset tool. I picked the GeoEye-1 Raster Type and clicked the Properties button.
I checked Merge Items on the General tab and clicked the Properties tab. I selected the DEM in the Orthorectification using elevation section and browsed to a DEM. Then I closed the dialog box.
I draged the North folder onto the Source box on the tool’s dialog box. Then I expanded the Advanced Option section and checked the options to build the raster pyramids and calculate statistics. Finally, I clicked OK and the first 4 images are added as one item to the mosaic dataset. Then I did the same for the South folder. This resulted in two image items in the mosaic dataset.
Next, I wanted to remove the black bits. There are only 2 items, so I opened the Attribute Table and added the Mask function to each item.
To do this, I clicked on the <Raster> field item in each row to get a little button to appear. Clicking the button opens the properties for the item. I clicked the Functions tab, then right-clicked the Merge Rasters Function and inserted the Mask function.
In the Mask function, I set the NoData Interpretation value to All – so all pixels must be the same to mask it out–and entered 0 for each band.
I clicked Apply to apply the addition of the Mask function, then added the Mask function to the second item in the Attribute Table and finished by closing all the dialog boxes and the table.
The color correction requires statistics to be calculated for each item. So in the Catalog window, I right-clicked the mosaic dataset and clicked Modify > Build Item Pyramids And Statistics. I unchecked Build Pyramids and Skip Existing and left Calculate Statistics checked and clicked OK.
Next, I opened the Customize dialog box and added the Mosaic Color Correction button to my toolbar.
I clicked the Mosaic Color Correction button to open the window. I selected my mosaic dataset and I started to apply different options for color balancing until I came up with one I like best. In this case I ignored all the other options and ended up with Balancing = Dodging, and Surface = Second Order.
This gave me the following result.
As I zoomed in on the edge between the two images I saw a slight difference, this was mainly due to the different dates and times of year for each acquisition. I ended up generating seamlines to make the edge between the two images look less apparent.
Now I have a great looking image to use as a base map for the rest of my work in this area.
