What's new
Photoshop Gurus Forum

Welcome to Photoshop Gurus forum. Register a free account today to become a member! It's completely free. Once signed in, you'll enjoy an ad-free experience and be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!

Calculating pixel between two different images PS. Is it even possible?


mystvearn

New Member
Messages
4
Likes
0
Hi,

I'll try making this as simple as I can.

First problem:
Lets say I am using PS with Intuos stylus to trace the picture of leaves from a same tree. Each image is traced and kept onto a different file. I then superimpose the leave tracings to see if the images match. I notice there are slight difference (as no identical leaf exists). Is there a way for me to measure that pixelated difference in PS?

Am I looking at a totally different software here?

Second problem:
Lets say I make a database of leaves from the same tree. Someone gives me a digital image of an unknown leaf and I need to identify if its the same species of tree as my collection. I know that the image resolution of the given image and my image database is different. Is it possible to measure this pixel differences?

The only way I see is for me to superimpose both images on a different layer and take measurements from that layer after saving it lets say jpeg, discarding the original source of both images as I can measure the relative difference of the pixels that way. That way I can tell if its both similar or not.

(Sorry for the green text, I can't see anything with the default black font on FF. Maybe my browser settings are different or I mistakenly made the font black).
 
Yes a subtraction or some sort of percentage, but how is it done?

I forgot to mention that lets say I trace each leaf with a different colour, how do I subtract the colour difference and not the background?
 
my final year project for my undergraduate degree was on something similar - shape recognition techniques and the morphology of Au nano-crystals - I don't think PS is what your after for a quantitative measurement - saying that, you could use difference layers to highlight differences between two tracings if they're of the same rough size.
 
my final year project for my undergraduate degree was on something similar - shape recognition techniques and the morphology of Au nano-crystals - I don't think PS is what your after for a quantitative measurement - saying that, you could use difference layers to highlight differences between two tracings if they're of the same rough size.

Is there a commercial software? Or is this more into computer recognition science here?

What I am trying to achieve is give some sort of measurement to a similar class of leaf until I find some sort of cut off point to say that it may not be from the same species even though looks the same.
 
yeah that does sound alot like what I was doing - bear in mind, this was 11 years ago, so stuff like this could well be commercially available now (hell maybe there is a ps plug in to do it now)

the gist of it was this - if you think of any 'shape' (like the outline of a particle or a leaf) one way to describe it is as a series of turns of 'n' degrees a distance 'x' away from a starting point as you work your way around the shape - if you plot x against n, you get what can be thought of as a waveform that describes the shape - from that you can use Fourier analysis to get a set of coefficients that describe the shape. If you were to do that for a set of leaves from a species that you know - you could get a sort of average descriptor of that species leaves - then comparing those to those for an unknown leaf you would be able to say if it was similar or not.

not sure exactly what level you're working on to know if that relevant to you or not - if you look up the work of a guy called JK Beddows, you can find more info - its geared towards metallurgy/material science - but the concepts should certainly apply to your case.
 
Oh, you mean morphometrics? I am doing postgraduate now. I know morphometrics work for my kind of work but then I need to manually put in the datapoints using the stylus
 

Back
Top