Subtract Background or Normalize ( IABgSub )

Parameters:

Name	Short Name	Type	Description
Background type	type	string	What kind of background to subtract. Calculated for each frame or for the whole image series.
Value	value	real
Background image	bgimagefilename	string	Filename, if image type is selected.
Background stage position	bgstage	integer	Which stage position within the background image set to be used.
Background frame	bgframe	integer	Which frame of the background image series to use
Average all background frames	avgbgframes	boolean	Whether to use all frames of the background image series
Shading/blank Image	shimagefilename	string	If image type selected
Shading/blank stage position	shstage	integer	Which stage position within the Shading/Blank image set to be used.
Shading/blank frame	shframe	integer	Which frame of the background image series to use
Average all shading/blank frames	avgshframes	boolean	Whether to use all frames of the background image series

Description:

Subtracts background. Typically use 1-15 percentile per frame, or selected ROI.
Percentiles are calculated per frame, per image series or based on the histogram of the histogram reference image.
Normalization divides the image by the reference image or by the mean of the selected ROI, the latter frame by frame.
Background or shading reference images cannot be loaded directly from multi-dimensional experiments. In such case subtract background image as individual images from each channel.
Matches channels when possible.
ADAPTIVE: (locally adaptive) tries to remove calculated background, but keeps darker areas above zero by decreasing locally subtracted background. Use high >90 percentile values for aggressive background removal.
Normalization by Image: first subtracts background image (alternatively omit background image name and give background offset as "Value") and then divides by shading image minus the background image
Optical density by Image: subtracts dark current (background) image (alternatively omit background image name and give background offset as "Value") from image and from blank image and then calculates LOG(I0/I).
Normalization by ROI: divides each image by ROI average and multiplies by the mean calculated for the whole image series.
Reference images: instead of giving file locations, reference images can be designated in the Image Window context menu. The first frame of the first reference window with matching dimensions and channel value will be used.
Normalization by Reference Image: use a dark current and a blank image for shading correction. A numeric dark current value may be used instead of dark current reference image.
Normalization and background subtraction by Background Reference Image: inhomogeneity of the background will be used as a bacground and then for shading correction. A numeric dark current value must be set for this.
Mean of pixels below percentile of max projection: A copy of the image series is maximum intensity projected, and then 2D maximum filtered (width=5) and the histogram is calculated.
For each frame the mean of those pixels will be subtracted which are below the given percentile of this projection image. Therefore, the background is determined over the same pixels for each frame. Use this method instead of percentiles per frame if the large number of bright details appear or disappear during the time lapse.
Median of pixels below percentile of max projection: similar as above, but interpolated median is calculated.
Note: percentile-based background subtraction may become noisy if too small histogram size (<1024 bins) is set in the Preferences+
Note: image and channel registration using Linear or Cubic interpolation in time lapses may inject noise in some background subtraction algorithms. Use image and channel registration with Nearest Neighbor interpolation in such case.