Image stitching lets you combine multiple image tiles to create a single high-resolution image that is beyond the normal aspect ratio and resolution of a microscope's field of view. For best results, the acquired image tiles should have overlaps with distinctive features that can be used as reference points for aligning tiles accurately.
Image Stitching Challenges
Sometimes, aligning images based on their image content is not always easy or even possible. For example, when two overlapping regions have very little content it can be almost impossible to tell how they should be positioned. Or when the overlapping region depicts a structure like a straight line that is translationally invariant it may not be possible to automatically derive a precise position of the overlapping images. In such cases, you may need to work on sub-sections of the tiled image mosaic and then stitched the aligned sections together.
You should also note that poor image quality, such as low contrast or the presence of excessive noise, can adversely affect the performance of the automated alignment methods. For such cases, you may have to process your images accordingly before attempting any stitching operation (see Image Filtering). Pre-processing may also be required if the brightness and contrast of the unique image tiles varies within your dataset.
The Layout Editor appears in the Image Stitching context.
Select the grid that corresponds to the type of collection used to acquire the image data (see Choosing Image Grid Layouts).
Load the configuration file, as described in the topic Loading Configuration Files.
In cases with good image content it may be possible to stitch all the image tiles sequentially. In other cases it may be best to stitch images tiles pairwise.
For complex cases with varying image content, it may be best to work on subsections of the data and then stitch the aligned groups together. You can move selected tiles with the Move tools (see Translating and Rotating Objects).
If required, you can apply an additional automatic alignment or use the Value Stitcher and Move tools to refine a selection.
If there are obvious seams between the aligned image tiles, you can blend the overlapping region so that the stitched result appears seamless (see Blend Options).
A number of configured actions and keyboard shortcuts are available for working with Dragonfly's Image Stitcher. The default settings for these actions are listed in the following table.
|
Action |
State |
Key |
Mouse |
|---|---|---|---|
|
Select channels (Reference) |
stateRefSelectionOne_OrsDatasetStitching |
|
Left mouse |
|
Select multiple channels (Reference) |
stateRefSelectionMultiple_OrsDatasetStitching |
Left Ctrl |
Left mouse |
|
Switch to multiple reference selection mode |
stateRefSelectionOne_OrsDatasetStitching |
Left Ctrl |
|
|
Switch to reference selection mode |
|
|
|
|
Select channels (To Stitch) |
stateToRegSelectionOne_ORSDatasetStitching |
|
Left mouse |
|
Select multiple channels (To Stitch) |
stateToRegSelectionMultiple_OrsDatasetStitching |
Left Ctrl |
Left mouse |
|
Switch to multiple To Stitch selection mode |
stateToRegSelectionOne_ORSDatasetStitching |
Left Ctrl |
|
|
Switch to To Stitch selection mode |
|||
|
Display next slice (in Image Stitcher) |
|
Left Ctrl |
Mouse wheel up |
|
Display previous slice (in Image Stitcher) |
|
Left Ctrl |
Mouse wheel down |