Detecting bubbles in the movie

3 views (last 30 days)
Vadoud
Vadoud on 13 Oct 2014
Commented: Vadoud on 13 Oct 2014
Hello Everybody,
I am trying to perform the PLIF technique on the multi-phase flow to measure the dye concentration field in the flow. For this matter, I have to detect and remove bubbles and their attributes in the collected images as contaminations of the desired data. I tried different image processing technique to detect bubble in a single image. Recently, I came across the Computer vision toolbox of the Matlab and its abilities for object detection. Here you can find a sample of my recorded images: https://www.dropbox.com/s/rnwr0o3w1w17wpl/bubble_moves.avi?dl=0 The dark moving round objects are bubbles and their associated shadows. I tried the multiObjectTracking demo to detect bubbles in this movie.
I really appreciate any help and advice that you can give me about my problem. Thanks
Vadoud
  1 Comment
Image Analyst
Image Analyst on 13 Oct 2014
Vadoud's "Answer" moved here.
Hi Image Analyst,
Here is the small size version of the movie:
https://www.dropbox.com/s/rnwr0o3w1w17wpl/bubble_moves.avi?dl=0

Sign in to comment.

Answers (1)

Image Analyst
Image Analyst on 13 Oct 2014
It's almost 900 MB. So I didn't download it. Either attach a single frame, or see what kinds of tracking and capabilities the Computer Vision System Toolbox has.
  2 Comments
Image Analyst
Image Analyst on 13 Oct 2014
Threshold for dark stuff, then filter based on the circularity or Bounding Box aspect ratio to get rid of the long black stripes. What's left should be bubbles, I guess.
I don't know what the PLIF technique is. There are some motion estimation methods that look over several frames to build like a "tube" over frames that can help predict where the object is moving to. I'm not too familiar with those algorithms.
Vadoud
Vadoud on 13 Oct 2014
Thanks for your response, I have tried all of these methods on my data. Unfortunately none of them worked. If you check the intensity of the bubbles, you will see that their intensity in some area is as bright as non-bubble areas. That is why threshholding does not work on my data.
And the PLIF is Planar Laser Induced Fluorescence. It is a non intrusive data acquisition method in fluids.

Sign in to comment.

Categories

Find more on C4ISR in Help Center and File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!