|
snow globe George from United States [3 posts] |
9 year
|
Gentlemen,
I am also new to your very cool product. Being Christmas season I would like to count various size particles in a snow globe. I have been playing with blob filter, blob count and blob size with some success. I thought while read these post to ask for your suggestion of an optimum method.
Thank you
George
|
|
|
Carl from Your Country [1447 posts] |
9 year
|
George,
It will really depend on the image. Can you post one here?
I'm not sure you can really count each flake but you can certainly quantify the density of flakes in the globe and how that changes over time. The reason is that flakes will overlap with each other and it will be difficult to isolate each individual flake with just a camera image. Using an overall estimate is the way to go.
STeven.
|
|
|
George from United States [3 posts] |
9 year
|
Hi Steven,
It is not actually a snow globe but it is the closest common example I can think of. I understand the over lap issue. But if we could count the different size blobs we could determine if there are too many large ones (>3mm) for example. I thought I saw an array of blob sizes, sometime reading before but didn't see that post again. I am curious how one would quantify density other than counting then dividing by volume, if there is a function to do this this might solve the problem too. The problemI am trying to solve is the see how fast the water clears . Please refer to attached image Thank you
|
|
|
Carl from Your Country [1447 posts] |
9 year
|
George,
The key words you mentioned are "see how fast the water clears" which makes this a relative problem rather than an absolute problem. For a relative problem you are just wanting to know when the density of particles gets low enough to declare it clear. That's easier than creating an absolute amount that is consistent across different scenarios.
What I would do is use the crop module to just isolate a part of the image that contains particles, then an edge detection module that would detect edges and then a Color Stats module after that to check the intensity level of the edges. Less edges means less particles. You would do this for an image full of particles and one without and see the relative change in intensity created by the edge module. That's the measure of how many particles are in the image.
If you have an unannoated unmodified original image of both one with particles and one without we can provide you with the above steps for testing. Note, the image you included seems to have already been processed in some way so using that would not create the right results.
Checking the amount of edginess in the image doesn't require knowing the actual particle count ... which isn't needed when you are comparing relative images. That makes it a much easier problem that actually counting the number of particles in the image. Instead we count the number of pixels and divide by the area which is how the average intensity value is calculated. A single pixels doesn't equate to a single particle but that's ok for comparison.
STeven.
|
|