Label Tracking Mobile Objects Labview

[tiho]

Hello!

I have a question using LabVIEW. I would like to track mobile objects. The mobile objects are spheres with equal parameters. I recognize them successfully. But I have the following problem. In the first image (focal plane 1) each one of the spheres has its own number. In the second image (focal plane 2) the number of the spheres increases by two. Then the spheres from the first plane will change their numbers or their labels. I would like to label or mark each one of the spheres on such way that each sphere will be always with the same number. How to do this? Please, help!

Regards

[Ben]

I think you will need to remember the sphere locations after the first frame. In successive frames, account for all of the spheres from frame 1 and then ID the new spheres.

"lather, rinse, repeat,"

Ben

[Bob_Schor]

If I handed you the Frame 1 picture with the five "spheres" numbered as shown, and then handed you Frame 2, without any numbers, how would you assign the numbers?  You have not provided any rational for assigning any numbering to Frame 2!  How do we know you aren't dealing with 7 totally different objects?

An obvious answer is that there is some "continuity rule/principle" that we don't know about -- either you have additional information about the spheres as a function of time, or as a function of space (you use the phrase "focal plane", which suggests "depth" to me).  I'd recommend that you get the "in-between" points and develop a "most-likely-correspondance" algorithm.  Write a little simulator program to test your algorithm ...

Bob Schor

[Ben]

---

@Bob_Schor wrote:

If I handed you the Frame 1 picture with the five "spheres" numbered as shown, and then handed you Frame 2, without any numbers, how would you assign the numbers?  You have not provided any rational for assigning any numbering to Frame 2!  How do we know you aren't dealing with 7 totally different objects?

...

Bob Schor

---

The current approach is doing a raster scan from top to bottom and assigning numbers.

The other factor ( that will be introduced) is that the sphere will most likely be moving (unless they are dead, see the image name).

In which case my outline above will have to determine the location of the spheres and then compute the distance between the sphere current location and the location from the previous image and assign numbers based on the minimal distances... That is more work than I want to do without a PO.

Ben

[Bob_Schor]

Ben,

     Is this your project?  You seem to have "extra information" (like "raster").  I find it interesting that we have two static representations and no information on how they differ (in time?  in space?) or the nature of the sampling (uniform spacing?  "chaotic"?  ordered?).  Your use of "nearest neighbor" has an air of "reasonableness", but without knowing something more about the "deltas" in the measurements, it would be a stretch to put much confidence in the conclusions from the analysis.

Bob "IMHO" Schor

[Ben]

---

@Bob_Schor wrote:

Ben,

     Is this your project?  You seem to have "extra information" (like "raster"). ...

Bob "IMHO" Schor

---

My Project?

No "That is more work than I want to do without a Purchase Order. "

I would have to be paid to develop that application since it would require quite a bit of work.

"Raster scan"

That was speculation based on the numbering in the two images. In each the numbers seem to be ascending as we work from top to bottom.

Based on the file name, I am guessing it will be microscope images and the rate of motion and count can be used to evaluate fertility.

Still speculating,

Ben

[tiho]

The spheres motion is chaotic. They differ in time and in space.