Align Tiled Series (Image Stabilizer) ( IAAlignTiledSeries )

Parameters:

Name	Short Name	Type	Description
Number of tiles in x	tilesx	integer	The image consists of this number of equal sized tiles in x dimension.
Number of tiles in y	tilesy	integer	The image consists of this number of equal sized tiles in y dimension.
Sub pixel accuracy	subpixel	boolean	If Yes then interpolated images are compared.
Magnification factor	magnification	real	The sub pixel precision is 1/magnification
Working image size, exponent of 2	workingsize	integer	Magnified images will be cropped to 2^size quadrangular. No crop if 0.
Difference Period	smooth	integer	Difference period gives frames how far from each other are compared
Interpolation type	type	string	Interpolation used for the transformation. Affects the smoothness of the result.
Compared to first frame	tofirstframe	boolean	Dislocations are calculated as compared to the first frame.
Reverse alignment	reverse	boolean	The last frame remains in its original position, not the first.
Cancel transformation if the shift is greater	cancelshift	real	Does not perform transformation if the absolute shift in pixels is larger than the given value. Zero disables this feature.

Description:

Align Series aka "Image Stabilizer"
Align Series registers (multiple) image series in time, preventing shaking of the image.
The alignment is calculated based on Image A, and all selected "Image B"s are transformed in the same way.
Alignment is based on sub pixel cross correlation maxima calculated for each frame from the single image series selected in "A". When sub pixel accuracy is enabled, the middle of the image is magnified to quadrangular (hidden) image defined by the working size, and the detection of dislocation is performed in the magnified image. Then the original image series is transformed.
When "Compared to first frame" is Yes then the sequence will be aligned to the first frame. If No, then local dislocations between pairs (or sets of multiple images if "difference period" > 1) of images is calculated as a running differentiation along the whole time lapse.
Be mindful of this mask if using non mask-compatible functions such as 2D DFT, image registration and template matching. Use fill mask with zeros for such functions.
Interpolation type: Use Nearest Neighbor interpolation with masks or when "...pixels below percentile of max projection" type background subtraction is used after this transformation. Smooth interpolation types may introduce noise into histogram-based functions.
Tiled operation: each frame of the tile is aligned independently. The resulting gaps are filled with MASK values.