Histogram Programs

[usf]

Hey, im not familiar with any photography. In fact Im not even sure how good my camera is. However, Im an engineering student and my professor wants me to take a histogram of stills of a video. My project requires that I characterize different behavior of plasma. And the most inexpensive way to do this is to take the histogram. But I have no idea what programs would do this.

So my question is what is the cheapest program that gives me simply a decent histogram at a given point? Are there any free trials of programs that i can download and use?

Thank you guys very much for your time.

[JonathanC]

Histogram of what exactly? If the plasma brightness varies over time, then this is one thing, if its the colour, then its something else.

[Dave Humphries]

Histogram of what exactly? If the plasma brightness varies over time, then this is one thing, if its the colour, then its something else.

Hi Jonathan,

I suspect both

Hi usf,

Welcome to the CiC forums.

Depending how many frames we are considering, and how many times you need to repeat the analysis, it may be possible to save still frame images from each video frame, then analyse these one-by-one.

I think I am correct in saying even something free like Picasa 3 (google it) has a histogram display from still pics. You'd want one that shows the colours separately I suspect.

I think we need more info. to advise better as Jonathan says.

I bet this is something Steaphany would know a bit about!

Perhaps we can get you "into" photography while you're here

Cheers,

[usf]

Well, I think we just need the breakdown of the colors. How much red blue and yellow in each section of the frame. Also, my professor said something about the camera and wether the waves would be affected. Anyone know what spec on the camera i need to refer to? I have a Canon powershot SD770 IS. 10.0 megapixels 3x optical zoom.

Actually, I am fascinated by the pictures posted on the site. I was browsing last night and I am in awe of all of your work. Not to mention that I have no idea what makes a good picture nor all the effort that goes into it. I commend all of you!!! Thank you for taking awesome pictures for us to enjoy. And I just might venture into photography as soon as i graduate .


Thank you guys for your time!

[Dave Humphries]

Hi usf,

Hmmm, vidoe and stills can be quite different to analyse like this...

If a still picture is viewed in an image editing program, most of these have a feature where they show the values of Red, Green and Blue within a pixel. Moving the cursor (via mouse) around the screen would allow you to note these down. That may be enough for you.

However, if the source is video, that is often times seen recorded as values of "Cr", "Cb" and "Y" and I wonder if this is why you said 'yellow' above - possibly mis-interpreting what the lecturer was saying? For the record, the "Y" is actually the luminance, or monochrome brightness, of the picture, not yellow. Also the Cr and Cb are what are known as colour difference signals, but I don't want to get bogged down in explaining that lot if I don't need to

Video can be converted to RGB values, or stills can be converted to Y,Cr,Cb (but often called LAB) if need be.

So, will you be taking the plasma pics with the SD770 - I'm not sure where that fits into things?

I'm not sure I'm helping much, mainly because I am still quite unclear what you are trying to analyse, and how.

Is there any chance of you posting an image?

Cheers,

[cneedha]

usf,
I think you have all the engineers in the audience interested now! It sounds like you are trying to quantify color differences in a plasma discharge- posting a sample image would help the experts (Dave et al) help you through the nuts and bolts of the data collection.
good luck,
chris

[wjh31]

I think something to consider, is that a picture of plasma will likely contain alot of things other than just plasma, and a histogram would therefore also be analysing whatever else is around, so you would either need to crop to just the arc manually or through software, or alternatively get the background fixed, e.g make sure the background of the picture is entirely covered with white card or similar for uniformity across the experiment.

I believe there are various ways to extract stills from video, including a way to extract all frames, though its not something i do often so couldnt refer you to anything off the top of my head. With all frames extracted you could then pass the lot through some histogram software to put out the histogram over time, with the assumtion that some parameters of the plasma are being varied with time, so therefore you can get to the variation of the histogram with the parameters.

It might be most efficient to write the software for the histograms yourself. With the image in RGB, each pixel with have a value of 0-255 for each RGB colour. The software just needs to go through and for each pixel, add 1 to the histogram for the corresponding value. I.E a pixel might have RGB value (137,143,23), so the software would add 1 to the 137 bin for red, 1 to the 143 bin for green etc repeated for all pixels. Your video is 640x480@30fps, giving 300k pixels per frame, so 9M points per colour per second. Thats a sort of range that your average computer could work through without much difficulty, especially if you dont want to take every frame, althought doing so might help reduce errors a-la 1/sqrt(n). leaving you just to worry about systematics rather than needing to consider random errors.

On thinking about it, using a white background might help get more data rather than cropping, as the white will reflect the colour of the plasma for you to measure aslong as you can lock the colour balance to prevent the camera skewing results.

You may get issues if the plasma is particularly bright, the core of the plasma may be so bright as to just show as white (aka blown highlights), so loosing usefull information at the centre, this would mean youd have to look at the edge of the plasma, and/or at reflected light to get a good reading.

Hopefully some of this is useful, without knowing particulars it could all be useless to you however...

P.S, my background is in physics; its not quite engineering, but hopfully close enough to be helpfull.

[KentDub]

So my question is what is the cheapest program that gives me simply a decent histogram at a given point? Are there any free trials of programs that i can download and use?

The tool you are looking for is Adobe AfterEffects CS4. It is "Photoshop for Video". Trial versions are avaliable

I wouldn't recommend wasting time extracting frames and running them through traditional image tools. You will be far more productive using AfterEffects. It will give you a histogram, adjustment layers, and most of what photoshop offers. In addition, it offers virtual-versions of many expensive peices of hardware used to analyze video signals, which may prove useful to your assignment. It is a daunting program the first few times you open it up, but feel free to post questions or send me a PM.

[usf]

Hi all,
I hope you guys had a happy thanksgiving!!

Thank you very much for your assistance. Well, I will be taking the video with my camera the canon SD. And I just discovered that the resolution of the still frames of a video are significantly lower than the images . Which means, that i will have to re-do all of my experiments, or hope that the department of engineers doesn't understand what i did with the camera and lets it slide.

As for what I am doing with the plasma: it is my final project (presentation due on Friday the 4th) and i have to present a poster on my findings. I have to characterize different aspects of the plasma. What happens to the intensity of the plasma as the pressure increases and decreases? What happens to the plasma as time increases, this includes the color (which tells us whats occurring with the energy), and the intensity of the light. That being said, I have used windows movie maker to get the stills and I have a free trial of Adobe Photoshop that im using for the histograms. I will be working on this tonight and hopefully have some graphs to show my professors on Monday.

http://www.youtube.com/watch?v=kLeXL570L2Y