How can you detect straight lines in IMAQ? There's no hough transform implementation.

[andru]

Hi,
Im doing a project in Labview/IMAQ, but C/C++ can also be. I need to know how can it be done most effectively. I mean, It works in IMAQ in LabView, just by detecting particles, and measuring their max intercept and elongation.But there is a problem of right thresholding the original(grayscale) image. I also looked into hough transform - the demos are impressive. But there seems to be no implementation of it in Labview/IMAQ. Can you help somehow? Maybe you know where in WWW I can find some good and relatively simple Hough Transform implementation in C/Labview? Or some other way to do it. The calculation should take under a second for 2.4GHz computer - and image is big.
Thanks in advanc
e.
Andru (CZ)

[Dustin_W]

Andru-

There is not an implementation of the Hough Transform for LabVIEW or C/C++ for IMAQ. You should be able to find some C implementations on Google, but it will likely require converting your image to a 2D array or calling a DLL.

[Andrey_Dmitriev]

Hi,

You can use OpenCV library:
http://www.intel.com/research/mrl/research/opencv/overview.htm

Hough trasform is inside (with source code).
You needs to write only small wrapper for calling functions from CV (not very complicated)

[andru]

Hi, guys!
DustinW:
thanks. I found some algorithm and used a formula node in labview and transformed my image to array, it works, but is too slow still. the image is 8bit grayscale but very big 3000x 3000 almost. i dont want to resample it or threshold, because i could lose some lines i need to detect. maybe reducing the number of angles a pixel votes for into the accumulator in the Hough transform will work. do you know this algorithm, maybe have some ideas? But the problem is the lines(scratches) i have are 10-30 pixel wide, and about 100-600 long. detecting them is not so easy - noise amount on the images is different. thanks for help 1nce more,
andru
Dymitr:
i will look into the CV...