Problem detecting multiple edges with a IMAQ Find Edge

[apatisk]

I'm trying to use the IMAQ Find Edge VI to find multiple edges in an image. The help file states that one can specify the number of edges to find, but my program is only able to detect the outermost edge even if I i increase the Line fit options -> Number of Lines.

Has anyone else experiensed problems finding multiple edges?

Does any other properties than Number of Lines need to be changed?

Attaced is an image of the program where I try to use two find edge blocks to find 4 vertical and 2 horisontal lines.

[muks]

You dont have the option to findall the edges along the lines in this function. You do have it in the edge detction functionthough. For this you can try extracting individual blocks if it will be helpful

[apatisk]

Thanks for the quick answer, muks!

The EdgeDetection VI finds all the edges, but it does not completely solve my problem.

My original plan was to use the "Straight Edges" output from the Find Edge VI to calculate the coordinates for the corners of the white boxes. (I need them to calibrate the image later on, 'cause in the system I'm developing, I will not know the position/distance/distortion of the white objects. The uncertanty of the objects position makes it difficoult to use multiple Find Edge VIs)

EdgeDetection does not provide any output of what edges/lines it finds, so I need an other way to find these corners.

It seems like IMAQ Extract Curves will give me the some of the coordinates og the edge, if i apply it after the EdgeDetection, but it will take some work to separate the different edges/lines or find the corner of the boxes.

Do you know any simpler way to find the edges EdgeDetection locates in a format sililar to Find Edges' Straig Edges output?

Or gor any Ideas for a good way to find the corners of the white boxes when the image is a bit distorted?

Thanks for any ideas! xD

[+39 0376 263525]

Hello Apatisk,

Have you thought about the use of a telecentric lens for the application? It gives a 2D view and will show the exact edge of the object, assures nearly zero distortion and can solve the problem of the exact positioning of the object thanks to its large field depth.

Alena Verameyeva

Opto Engineering, Italy

alena.verameyeva@opto-engineering.com

[fink]

Okay yes telecentric lenses will help for the optical quality of the measurement, and furthermore if you are also using a telecentric illuminator in front. the problem then is to have a camera with a short exposure time to avoid overexposition.

For the LV processing, you can also use a rake (from different sides and orientations, because you can have "shadow" effects) and then to sort your edges results afterwards by analyzing their position with classical array analysis (it is now part of the edgetool boxes but before you had to do it so). Enter in the vi diagram and check the kind of wires you get. You should need to unbundle clusters of arrays and so...

I did this to find contours of concave objects.

hope this helps

F