Tracking movement of bubbles in water to obtain velocity

Hello,

I'm working on a project that involves tracking bubbles moving through a water tunnel and using the information to determine their velocities. I'm very new to Matlab and I'm looking for a little guidance on where to start.

I have the basic toolboxes that come with the Student Suite, and Computer Vision; just going through the examples, I found numerous ways to track the bubbles, but I don't know how to convert the tracking to velocity. Any help you can offer is greatly appreciated!

Thanks in advance,

Rob

Answer 1

[Image Analyst]

Robert: To get "real world" units you need to calibrate so that you know how many pixels correspond to 1 cm (or whatever). See my attached spatial calibration demo.

FAQ: http://matlab.wikia.com/wiki/FAQ#How_do_I_measure_a_distance_or_area_in_real_world_units_instead_of_in_pixels.3F

In your reply to Sean you say you need "information about the centroids' location" - what does that mean? You have the x and y location of the centroid - what other info do you need about the centroid location other than the coordinates?

[Robert]

Thanks for the demo, I'll give it a look asap! By the information I meant precisely the x and y coordinates; let me rephrase, I've gotten to the point where (going by the examples) I can overlay the centroids in a plot and confirm that their locations are correct , but I don't know how to use this to get their x-y position.

[Image Analyst]

That doesn't make sense. You said "I managed to find the centroids" - that means you have the x and y locations. In fact you said you plotted them over the image doing something like

measurements = regionprops(labeledImage, 'Centroid');
centroids = [measurements.Centroid];
xCentroids = centroids(:, 1);
yCentroids = centroids(:, 2);
plot(xCentroids, yCentroids, 'r+', 'MarkerSize', 2, 'LineWidth', 3);

So again, you already have the x-y position and there is nothing else left to get. You can attach your image and m-file if you want.

[Robert]

I'm sorry, you're right, I did have the positions already; I thought I had to use another function to get their position, my (embarrassing) mistake. Thank you for your help and your patience!

Rob

Answer 2

[Sean de Wolski]

Easiest way would be to find their centroids ( regionprops or vision.BlobAnalysis), calculate the distance between them ( hypot) and divide that by the time step.

This will give you pixels/second, if you know the pixel size, multiply by this.

[Image Analyst]

Robert's "Answer" moved here since it's a reply to Sean:

Sean,

Thanks, I followed your recommendation and I managed to find the centroids but now I'm just having difficulty getting the information about the centroids' location; I thought I would want to use "getpixelposition" but that doesn't seem to be working (probably for an obvious reason I don't know yet lol).

Do you know of a better function that I should use? And thank you for getting me started with the centroids!