Maybe it is easy problem to you but it's urgent to me.

[Swedlin@hotmail.com]

I am still new in machine vision. I have successed with IMAQ Builder 6 to filter out the objects interested from a image and the mid-result shown as the attached image. Some fibers(very small and tiny fibers) are the objects interested and shown as the attached image. The main task right now is to find out
the longest one of fibers in the image and recognize the both endings of the longest fiber found in order to further move it to a position desired.

I have to implement this algorithm but it is quite hard for me to set up this algorithm now because I am still lack of image processing.

I will set up my own iamging system and get a serious images that are quit similar to the one shown as the attached image
. Of cause all appearances of the every fiber are diffirent but they are quite similar under our human eyes.

Maybe this is trivial for some of you and I'd like to have your solution as reference whatever processing package you used. Of cause IMAQ Builder 6 is best for me to try again. Thank you very much for your time and energy, any comments and your solutions!

Cheer!
Swedlin

[BruceAmmons]

Under binary image analysis in Vision Builder, select particle analysis. The property you want is max intercept. This gives you the longest line in the particle in any direction. Once you identify your particle, you can pull out information like orientation, center X and Y, etc. Using the center position, orientation and length, you should be able to calculate the endpoints fairly accurately.

The main problem you could have is when fibers are touching or crossing each other. They will be treated as a single fiber, which probably isn't what you want. It would take much more advanced algorithms to fix that, though.

Bruce

[Swedlin@hotmail.com]

Thank you very much, Bruce!

Your explaination highly enlighten my idea to the problem I have faced. I have learnt some thing more from your suggestion. I have got the longest fiber from the image I processed. I will further figure out what you have meant in your sugguestion.

The main problme as you mentioned I have forseen and I have to use other measures to solve it (for example using the output of the strength of fiber).

Thanks again!

Swedlin