Integration field for OME-XML

[crueden-x]

The acquisition systems at LOCI have a configurable parameter for "integration" or "averaging" that controls the number of sequential frames that get averages, to improve the signal-to-noise ratio.

This is very similar to the Binning attribute of DetectorSettings, except that it is an average over T rather than over XY.

From the book "Video Microscopy: The Fundamentals" by S. Inoué and K. R. Spring:

"Because noise is random and the signal is not, a running average both reduces the noise contribution and enhances the signal component of the output digital image. ... A running average is a very convenient way to monitor dynamic events under low-light-level conditions, as in studies of fluorescently labeled living cells. Image quality is improved by averaging, with some sacrifice of temporal resolution, but with the benefit of a constant video output that allows one to monitor specimen focus and position."

Thus, I suggest adding an "Integration" attribute to the DetectorSettings element, with type="PositiveInt" and use="optional".

[crueden-x]

Another useful explanation from "Principles and Practice of Variable Pressure/Environmental Scanning Electron Microscopy" by D. Stokes:

"2.6.8.1 Integration

"Live digital images can be placed in a framestore and then integrated to form a single image. The process of integration smoothes out noise by adding a specified number of successive frames together (2, 4, 8, 16, 32, 64, etc). Constant signal detail is added together, while random fluctuations (noise) tend to cancel out. Integration is usually performed at fast frame rates and/or small beam currents, which is also an excellent way of avoiding artefacts arising from charge accumulation at slow frame rates and/or high beam currents. Note that integration is not useful if either the sample or the beam is drifting, since this causes image blurring.

"2.6.8.2 Averaging

"Also known as a recursive filter, this method increases the signal-to-noise ratio by giving a running average of several frames in real time, with pixels weighted so that the most recent frame contributes the most. Unlike integration, averaging does not stop after a specified number of frames, i.e. it is a continuous, real-time process, which means that contrast and/or brightness can be adjusted in real time."

[ajpatterson]

Changes in ​https://github.com/openmicroscopy/bioformats/pull/424

[Andrew J Patterson <workonly@…>]

(In [31173eee6358c0c6ad3631f193d780e482178aa0/ome.git] on branch develop) Adding Integration field, see #8599