1
$\begingroup$

I am playing with a weird dataset of ternary images (+1,0,-1 values only) which happen to be very sparse (avg. > 90%). I would like to determine the most relevant "islands" (or should I call them blobs) of non-zero elements.

Algorithm complexity is a key factor since I would like (hopefully) to deploy it in real-time over frames acquired by an RGB camera.

Any suggestion/pointer would help.

Thanks!

Improve this question
$\endgroup$
3
  • 1
    $\begingroup$ How do you define "relevant"? You might look for large connected components (say for nearest-neighbours, or nearest and next-nearest). $\endgroup$ Commented Mar 6, 2019 at 18:44
  • $\begingroup$ With "relevant" I meant the first N islands as sorted by size (i.e. number of non-zero elements). At the moment I am working on a simple algorithm that recursively partitions the focal plane and counts the non-zero elements combined with some heuristics (thresholds) I extrapolated from the dataset. However it is not general and might not be performant enough. Thanks for your inputs, i will post the solution if I find (an efficient) one. $\endgroup$ Commented Mar 7, 2019 at 9:14
  • 1
    $\begingroup$ The standard algorithms for finding connected components in a graph use either breadth-first search or depth-first search, and are $O(V+E)$ for a graph with $V$ vertices and $E$ edges. $\endgroup$ Commented Mar 7, 2019 at 12:35

0

Reset to default

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.