Imagine a video recording of a bunsen burner flame.

We could implement a crop that displayed just the flame, or part of the flame.
Beneath, we would have the audio wave timeline.

Is it possible to average the colour on each frame, and place that number onto a colour wave time line?
This would chart the change in colour, and from this we could get a detailed image of the temperature change, as air is added to the flame.

Perhaps the histogram tools might be used?

Clearly(!) the eye can see the colour changing, but having a readout would at least standardise the changes occurring.

If this is not possible....... has anybody come across such a 'video colour analysis program?

MATLAB.

Thanks for that lead.
MATLAB is clearly a massive project.

I spent a good hour working through the information.
Most of my effort went into reading through 'Image processing & computer vision'.

Drilling down, in 'Concepts' I found this:

Image Pixel Values and Image Statistics
http://www.mathworks.co.uk/help/images/ ... #f11-13085

Half way down the page is:
RGB Image with Line Segment Drawn with improfile
Easily spotted, as it displays a bright picture of 'peppers'
'improfile' refers to a function that is described in the previous paragraphs.

The text always refers to 'image' .... and the corresponding data per image is substantial.

I noted, at the bottom of the page:
Image Mean, Standard Deviation, and Correlation Coefficent
You can compute standard statistics of an image using the mean2, std2, and corr2 functions.
******

This appears to be an enormous collection of software....... only that it is so vast; there is no pricing!

You have to speak to a member of staff, before even having a clue as to the sort of money that is wanted, and one is even warned about needing to setup an account, before the dreaded 'sale rep' contacts you.

It's definitely weird...... why do they want my account details before even telling me the prices.
Does anybody understand the MATLAB business?

It doesn't look as though this is relevant to people with web cams and mobile phones.
I'm guessing it's for people with big research grants, or corporation R&D departments (and students).

Does anybody have any experience with them?

Used it during my time at the university. Simulink was a heaven's send for creating control designs. Yes, it's heavy, it's versatile, ... it's expensive. But then if you are into image processing then, I'm afraid, the forum for a nonlinear video editor may not be exactly the best place to go.

Yes I hear you.

I do have a gut feeling that what I'm thinking of, is probably not too far from what has already been designed.
In fact, many of the produced tools are far more complex, often related to RGB analysis that then allows tweaking of the colour mix.

My requirement is purely for an assessment of the actual colour of the flame - probably less than a thousandth of what MATLAB offers.
The averaging of the flame colour pixels, purely to provide a more accurate single point output of a given area of the flame.

Other uses might be in fabric, where multiple coloured threads are woven.
The individual thread colour changes are irrelevant to the created effect, given different exposure to light.

Either way, it would appear that 'this' area of image examination has not achieve 'standard' status.
What a bugger eh?

I'll keep my eyes open