LabelMarkers function - any way around the MaxLabel = 255 limit?

[losh96]

Hello,

I am using the LabelMarkers function(IPP 5.2 beta) as a image "blob" finder and it has worked great except for the fact that you can only distinguish up to 255 blobs. After 255it resets marking to 1 so you end up with multiple blobs with the same marking.

Are there any efficient ways toget around this? I am planningon submittinga suggestion to Premier Support to possibly have the function return a 16u marker image, but I wanted to see if you had any suggestions to work around this problem.

Thank you for your time,

Milos

[Intel_C_Intel]

Hi, Milos,

You are right, the existing segmentation functions in IPP use 8u data for marker values ie labels from 1 to 255 are possible. Even more, 255 or 254 values are used in segmentation functions so the recommended values are from 1 to 253. If an image contains more markers they labels are repeated from minLabel to maxLabel.

We are going to add segmentation functionality using 16umarkers. Your submitting of this issue to premier site will help us in it.

What do you think on segmentation of 16u and 32f images?

Thanks,

Alexander

[losh96]

Alexander,

Thanks for your reply.It would be great if you added segmentation functionality using 16u and 32fmarkers, especially in the 16u case since it will increase the number of labels dramatically. I currently don't have a need to segment 32f images, but I am sure there are people out there to whom that might be useful.

Is the 16u functionalitysomething that would be available with the final version of IPP 5.2?

Milos

[Vladimir_Dudnik]

Hi Milos,

unfortunately functionality of IPP 5.2 already frozen, we are at the middle stages of preparing release, so we only can add this functionality after IPP 5.2 release.

Regards,
Vladimir

[apolo74]

Dear Milos and Alexander,
I would appreciate very much if you give me a hand with LabelMarkers. Like you, I also need to analyse (quantity, areas) the different blobs I have in my current image but don't know how. I tried to do this with openCV function cvFindContours, but the blobs disappear and instead I only get fuzzy white points in the borders of what used to be real blobs... I couldn't find out the cause of this, so I decided to analyse the blobs directly with IPP. Could you please give me an example of the steps and functions and their parameters to analyse blobs?
Thanks for any help you guys may give me,
Boris