How could I identify an object that has a Gaussian distribution and compute its area?

I'm researching different types of plasma discharge, and I have this imagethat I converted to grayscale and is basically light glow in a half Gaussian distribution in both its length and height. I want to identify (essentially accurately outlining) the area/boudary where the light has diminished 50% and 90%. Once I identifying that area, I want to compute its area. Would anyone have any idea how to do this?

Answer 1

[Raunak Gupta]

Hi,

I see a similar question answered here about fitting gaussian distribution to the intensity of an image. There, mixture of gaussian are assumed but for your application I think one gaussian distribution can fit the data. You may also want to look the documentation of fitgmdist for more information about how to input data into it. For single gaussian ditribution, select number of components as 1. You may draw the cluster lines as mentioned in the example for visualization of fitted gaussian distribution.

Hope this gives a head-start to solve the question.