Maximum Possible Resolution

[AndrewZ]

I would like to make the most accurate possible measurements at a distance of 2 meters. If it is possible, I would like to get measurements accurate to 100th of a mm or maybe to a 50th of a mm.

Is this just a pipe dream or is it possible?

What is the the best I can hope for? and at what price?


Currently I have a Sony XCD-X710 camera and I have been able to get about 10 pixels/mm at a distance of 180mm.

Thanks in advance,

Andrew Zwart
Test Engineering
Wescast

[Ken Pietrzak]

If I understand you correctly, you want to measure something that's about 2 meters in length. The approach here would be to use 2 cameras, one camera focussed on each edge. The field of view of the cameras and thereby the resolution would be limited to the tolerance of the object's dimension and the tolerance on the object's placement or fixturing. To calibrate the system you would have to place an object in the field of view with a known dimension.

[AndrewZ]

I only need to take measurements within an area of, say, 8cm by 8cm. The object of interest would be at a distance from the camera lens of about 2m.

I would like to know what kind of accuracies are reasonable to expect with a different lens for my camera or with a different camera.

Thanks,

Andrew

[Scott_R]

Andrew,

The resolution of the camera will tell you the smallest object that you will be able to measure in your image. For example, if your camera is 640 x 480, then you will have 640 columns of pixels and 480 rows. If the image area that you are analyzing corresponds to 64mm by 48mm, then you will have 10 pixels/mm in your image. You can achieve this same goal on a larger image area by using a camera with a greater resolution. In order to determine what lens you will need, you need to look at the working distance, the smallest object that you want to be able to discern in the image, and the size of the area under inspection.

To look at different cameras that you could use, I would suggest taking a look at the NI Camera Advisor, as it contains a list of cameras that have been tested and are known to work with the different NI framegrabber boards and the NI-IMAQ drivers.

Industrial Camera Advisor
www.ni.com/cameras

Regards,
Scott R.
Applications Engineer
National Instruments

[pachacuti]

The minimum feature that is resolvable is a funtion of the optics used. It can be approximately calculated by (2 * wavelength * F#) where F# is distance to the object divided by the lens diameter. If that is smaller than the projected pixel spacing of your camera, i.e., the field of view divided by the numbers of pixels, you still won't be able to 'resolve' it.

[AndrewZ]

Is that the wavelength of the colour of the object I am viewing with the camera in the formula?

Thanks,

Andrew

[pachacuti]

Sorry for not being clear. The wavelength is the color of the light used to form the image on the camera array. So it is the wavelength of the illumination assuming the object doesn't alter it. If you are using a green LED at 520 nm to illuminate your scene then it's 520 nm. If you are using white light, you have a range of wavelengths but using 550 nm will get you close.