Geomatica OrthoEngine - Mosaic

This section deals specifically with the Mosaic processing step within Geomatica OrthoEngine.  Topics include automatic mosaic, colour (radiometric) balance options, and unreferenced mosaics.

Automatic Mosaicing is leaving black holes in my mosaic - Why?

Can you tell me more about normalization in Automatic Mosaicking?

I have corrected imagery and would like to use OrthoEngine's mosaicking capabilities.  How do I mosaic my images?

The background colour in my final mosaic is Black, how do I change it to white?

Which methods are used to generate cutlines in OrthoEngine automatic mosaicking?

Why do I have a huge mosaic file?

Can I mosaic unreferenced imagery?

Q) Automatic Mosaicing is leaving black holes in my mosaic - Why?

A) Some scanned photos have areas of black, which have a brightness or digital number of 0. In this case, there is no data until a much brighter number, such as 80, is encountered. This is usually due to enhancements applied to the scanned image. The cutline generation algorithm compares the data values (80+) to the blank areas (0) and considers the blank areas to be background data. It then clips out these areas.

There are two solutions for this:

1) Recommended: Scan your photos carefully, using a good quality scanner - This will help to make sure that your range of values is reasonable. If you find the image too dark or too light and you are applying enhancements, make sure that you do not use tail-trimming. Tail-trimming will shift the histogram of most pixels, but leave the extremely dark or extremely light values set to 0 or 255. If your values are quite dark, start your enhancement by adding a bias to all pixels (i.e. add 25 to your image).

2) In a pinch: If you have already created your orthos, then they will already have a black background area. Adding a bias is not appropriate in this case. If you do not have any pixels with a value of 255, then use Replace Image/Photo Values to change all 0 pixels to 255. Run the automatic mosaic, then replace all 255 values with 0.

Back Top

Q) Can you tell me more about normalization in Automatic Mosaicking?

A) Normalization is used to even out the maximum and minimum effects in tone to achieve a more pleasing mosaic. This feature can be set up differently for each image that will be included in the mosaic by using he arrow beside the Normalization column. Alternatively, you can set the Normalization to be the same for all the images by selecting the feature in the Normalization list and clicking Apply to All.

- None will leave the images as is.

- Hot Spot will normalize the brightness over the image and will reduce the effects of solar reflections. It will however, not remove the spot reflections from lakes, cars and buildings.

- Across Image 1st Order is used to correct the gradual change in brightness from one side of the image to the other.

- Across Image 2nd Order is used to correct the gradual change from dark to bright to dark or vice versa across the image (this is also known as "antenna pattern."

- Across Image 3rd Order is used to correct gradual bright and dark patterns from one side of the image to the other.

The Help contains some more detailed information about Normalization.

Back Top

Q) I have corrected imagery and would like to use OrthoEngine's mosaicking capabilities.  How do I mosaic my images?

A) This is actually a fairly easy procedure to set up in OrthoEngine, but there are a few things to remember.

1) With OrthoEngine the imagery must have the same pixel size. If the pixel size is not consistent you will have to use the reproject panel to resample your imagery. Please see the FAQ "How do I resample to a different pixel size?" for more information.

2) Start OrthoEngine and select 'file new', set up a new project file (.prj) and ensure that 'Math Modeling Method' is set to 'None (Mosaic Only)'

3) Set up your projection, ensuring that the pixel size is consistent with the pixel size of your input imagery.

4) Under 'Processing Steps' select 'image input' and select the open image icon. Under the 'Open Image' panel select new image and select each image to be mosaicked in turn. This step is simply bringing your imagery online with your project file. After bringing all imagery online you can select the 'image layout' icon for a quick overview of the relational position of each image.

5) If you want your output format to be Geotiff or Tiff World choose Ortho/Mosaic Output Format under the 'options' menu. Otherwise, your output will be in .pix format.

6) You can now proceed to 'mosaic' under processing steps, define you mosaic area extents, and begin either a manual mosaic session or setup Orthoengine for Auto-Mosaicking. Try an Auto-Mosaic first as you may be able to avoid the extra work involved in the manual mosaic workflow!

Back Top

Q) The background colour in my final mosaic is Black, how do I change it to white?

A) Under the Utilities menu (this in on the main OrthoEngine panel), select 'Replace Image values', set your input file; your replace option to 'equal to', value to replace will be 0, and the value to replace it with will be 255.

Back Top

Q) Which methods are used to generate cutlines in OrthoEngine automatic mosaicking?

A) In the Automatic Mosaic procedure, PCI Geomatics software provides three options of generating cutlines in an overlapped area between two images:

1. Min Difference

2. Min Relative Difference

3. Edge Features

All the above methods are criteria for defining cost functions. The first is based on selecting the pixels with the minimum difference of grey values between two images as the cutline pixels. The second one is to select pixels with largest gradients as cutline pixels (e.g. edge of a road). The third method is a combination of method 1 and 2 so that they may compensate each other to obtain a better result. Users may experiment with these methods to obtain the best solution for their applications.

Back Top

Q) Why do I have a huge mosaic file?

A) This can sometimes happen if you haven't set the background value for the DEM file, particularly if you have a background value that is a very large negative. When OrthoEngine computes the output footprint of the ortho, it basically projects the original photo extents onto the DEM and uses the intersection of the photo and DEM as the footprint.

If you haven't told OrthoEngine what background values to ignore in your DEM, sometimes the intersection between the photo footprint and the DEM can occur at the background value (i.e. -999999). Obviously, extending the photo extents down this far scales the footprint to be very large. This is what causes the large black buffer around the ortho.

If you need to check the background value of your DEM, try opening it in a viewer (File | Image View), then clicking onto whatever black background areas you may have. The "Int:" value on the main OrthoEngine panel will show the background value.

Back Top

Q) Can I mosaic my unreferenced imagery?

A) In previous versions of Orthoengine it was required to have georeferencing information assigned to your input images.  In version 10 however, you are now able to mosaic unreferenced imagery.  The steps for this procedure are as follows:

• Start a new Polynomial project

• Select PIXEL as output and GCP projection

• Enter 1 for Output Pixel Spacing

• Add 2 or more images to project

• Collect GCPs on one image, using another image as geocoded reference

• Open Geometric Correction panel

• Move the image with GCPs to Images to Process

• Click Use Raw Image as Master for second image

• Click Correct Images

• Mosaic images (automatically or manually)

Refer to the tutorial for more information on this topic:

Mosaicing unreferenced imagery in OrthoEngine

Back Top