How Can I estimate the true width of a particle?

[Wilfredis]

 Hi, I'm trying to estimate the real width of an particle. I'm considering that its height is equal to the Feret's Diameter, but I can't find its width. What I want is to circumscribe a rotated rectangle (whose major side is aligned to the Feret's Diameter)  to each particle detected on an image. can anybody help me?

[Wilfredis]

Actually, also the axis of minimum moment of inertia (I can find its equation) is useful for me for aligning the rectangle. In both cases, I have not could estimate the width of the rectangle.

[Coamín]

Hi Wilfredis

I can recommend you check the Manual Concepts. In that manual has a lot information about the theory of your needs. Page 186

Check this link please.

http://digital.ni.com/manuals.nsf/websearch/55A90080787F932586256D3C0065F4DA

If you have more question, just tell me.

También puedes escribirnos en español!!!

Hola Wilfredis

Te sugiero checar el manual de conceptos de IMAQ. En ese manual viene bastante información al respecto y te podría sacar de dudas. Viene información en todo el manual, puedes checar particularmente la pag 186.

El link es el siguiente:

http://digital.ni.com/manuals.nsf/websearch/55A90080787F932586256D3C0065F4DA

Si tienes alguna pregunta solo escríbenos

Saludos

Coamín Cruz

AELatam

[JamieB]

I had a similar requirement and ended up doing the following:

-Extract the particle to another image

-Rotate it according to the oreintation property obtained in the first particle analysis

-Measure the particle in the rotated image and get the bounding box width and height

This has worked well enough for me, but I am generally dealing with small particles and on a very fast machine.  I was surprised that the NI particle analysis provides the max feret diameter, but not the min. (i.e. the length but not the width).  If anyone has a better method, I would be very interested.

Jamie

[comrade]

I'm currently stumped with the same problem of minimum Feret diameter. First I came up with the exact same procedure as JamieB. The problem is though that the rotation of the particle will modify its shape. This is neglectable with particles of reasonable size. But from a certain size downward the shape distortions become unacceptable, thus yielding erroneous diameter readings.

So, simple question: Has someone implemented a true minimum Feret diameter computation and is willing to share it?

[KevinC]

From the questions, I can't quite tell what is desired: the minimum Feret diameter (smallest possible caliper measurement), or the minimum width perpendicular to the max Feret diameter (bounding box with major axis length equal to the max Feret diameter).

Regardless, this little example shows how to calculate both without reinterpolating the image.

[comrade]

This seems to do the trick, as far as the min feret diameter is concerned. Thanks a lot for the quick  - and slick - answer.

BTW: Taking the height AND the width of the global rectangle in "Min Feret Bounding Box.vi" allows you to get the desired value within just above 90 degrees instead doing the whole 360 rotation.

[KevinC]

Good catch on only needing 90 degrees to find the min Feret diameter. I feel a little foolish to have missed it.  Anyway, I think that the general strategy is clear: you can use Mask To ROI along with Particle Analysis.

[comrade]

Yep, example was quite self-explaining. I knew the key was to rotate "the coordinate system" instead of the image (or particle), but the detour with the Mask2ROI and transforming the ROI is just clever. Thanks again.