01-07-2009 02:04 AM
01-07-2009 02:07 AM
Hi Aneece,
where do you have problems? Can you move your stage? Do you get your image? Do you want a 100% match (most improbable)? Do you know Vision?
Mike
01-07-2009 03:15 AM
01-07-2009 07:48 AM
Hi Mike,
thanks for your reply.
I can move my stages throug PCI- GPIB interface with labview, I really new to vision system, I would like to get capture predefind positon of moving object and save its image as a reference, then I will move it to another positon, afterwards I need to move back to the same first positon upon matchin the real time image with the reference one. your help will be highly apprecited. thnk
01-07-2009 07:57 AM
Hi Aneece,
maybe it would be easier to simply store the position of the moving stage and use counting motors , but i think it should also be possible with some image processing functions. You can create a greyscale image and calculate a image with only edges visible (i think this minimizes the color differences). If you have the edge window from both images (reference and real time), then you can subtract them from each other and stop your stage on the position with the minimum mistake.
Hope it helps.
Mike
01-07-2009 08:30 PM
Hi Mike,
one thing that I didn't mention to you that every time the homeposition of the moving stage will be different, so i cannot store the position otherwise is simple as you said. can you get an example of the solution that you give (greyscale image) some thing similar.
thanks
01-07-2009 11:42 PM
Hi Aneece,
how do you store the new/current home position. As i understand it, you have to store a new picture for a new home position, but why can't you store the stage position at this time?
Mike
01-08-2009 01:35 AM
I suppose the home position can't be fixed between runs because the system may have been operated for something else. So there is no garantee that the position has been maintained. I have the same problem with a microscope.
What I have done is
1 - to find the ratio between the positionning motor steps and the pixel size ;
2 - to define a reference position on the object, easily detected by image analysis and detection of specific details (line crossings, corner...) whith spatial coordinates (relative to the image coordinate system, accurately measured ;
3 - Then, for each new run, to manually position the object under the microscope, close to the reference position ;
4 - to detect the reference details on the new image, and subtract their position from their original position on the reference image. This gives the relative position of new image.
5 - to convert this DeltaXY into a move (number of X&Y steps, using the ratio discussed above)) and to re-home the system.
Hope this helps
01-08-2009 02:09 AM
thanks Mike/Chilly, and sorry if I confuse you, I will try to make it more clear, i have two objects need to be aligned (parallel to each other with specific distant) every time, the moving object (Ob1) will move backward a certain distance (this distance can be fixed) while the other object (Ob2) changing with new object(Ob2 changable one), so when I need to move Ob1 forward to the Ob2 the distant between 2 objects will not be the same as the previous one and that due to some misalignment, so that is why I need to picture the desired distant between 2objects and make it as reference, and then when i need to get to the same position while the real time camera capturing the moving stage until the image match the reference one the moving Ob1 will be stop. I really don't involve in any image processing before so i hope that i can get help on this.
thanks all