Image Quality metrics


Are there any already available image quality metrics plugins for ImageJ?


Hi, I think that the question is rather vague. Can you clarify what is meant by “quality” within this context… i can think of several different things that could affect the quality of an image.


I would like to have an image which evenly focused, crisp focused, evenly illuminated, optimally illuminated, even contrast etc. In short, parameters which can quantify these things will help me acquire better images.


I think you would need to break down the problem into smaller procedures that attempt to quantify all those requirements, including all those unknowns under "etc."
I do not think there is currently available code to do what you want.
If you think about it, without an underlying model of what the ideal image is, it would seem difficult to define a scale to benchmark against because you never know what you might expect in an image. The image might have uneven contrast due to the scene contents, not because of some artefact. Or might not have crispness if there aren’t many features in the image.


Check this out


One important quality problem with fluorescence microscopy is the photon count - which can be low. Part of the problem is that our exensive microscopes don’t count photons.
A useable measure for noise in static specimens is to compare two replicate images - same specimen 2 pictures, their difference is the noise.
The similarity between the replicates is clearly shown in a scatterplot which can be quantifies by measuring the correlation. We have converted this method into a scheme for making noise free measurements of colocalization.

Noise and Colocalization in Fluorescence Microscopy: Solving a Problem.
Microscopy & Analysis, Sept. 2008, 7-10.

Jeremy Adler