I want to manually measure the dimension of a pattern printed on a flat surface, with an accuracy of about 0.1mm. For simplicity sake, let's say this is a black rectangle, printed on a white background, and i want to measure its width and height. The dimension can be between 20cm and 80cm.

I have tried with a vernier caliper, but placing the caliper accuratly enough on the rectangle borders is impossible, especially since the caliper jaws do not lie flat on the surface.

What kind of measuring instrument could i use? Is there something dedicated to this task, or a way to use a standard instrument to do that?

  • $\begingroup$ You can lay a tape across it and take a couple of close up photos from either end. Then magnify the photos. $\endgroup$ – kamran Apr 7 '20 at 3:10
  • $\begingroup$ Scribing lines with a hardened point it usualky effective and with the use of engineer’s blue works well $\endgroup$ – Solar Mike Apr 7 '20 at 4:58
  • $\begingroup$ What are the range of dimensions to be measured? $\endgroup$ – Transistor Apr 7 '20 at 6:38
  • $\begingroup$ @Transistor between 20 and 80cm, i have updated the question. $\endgroup$ – Adrien Descamps Apr 7 '20 at 19:07
  • $\begingroup$ @SolarMike Not sure what you mean, can you specify? $\endgroup$ – Adrien Descamps Apr 7 '20 at 19:13

I had a plastic ruler at one time in my life and it looked something like Figure 1.

enter image description here Figure 1. Overview.

enter image description here

Figure 2. Close up of precision grid.

enter image description here

Figure 3. In-use.

How to use:

  • Align the right edge to the nearest whole cm less than the width of the item to be measured.
  • In the precision section, find the closest intersection between a sloping line and a horizontal line.
  • Add the three numbers. In this case it's 30 + 5 + 0.7 mm. = 35.7 mm.

I drew this in a few minutes on a CAD program. Vertical spacing is 2 mm. Print it off on laser-compatible transparent film on a decent laser printer and you should be in business. If you mirror the image before printing you will print on the "back" of the film and will have no parallax error when the film is pressed on the part. To make a longer scale you can tape several together.

  • $\begingroup$ That does not show as 2mm on my computer - more like 2.2mm according to the steel rule I have... $\endgroup$ – Solar Mike Apr 7 '20 at 18:33
  • $\begingroup$ Zoom tour browser out to 10/11 of the current resolution and you'll be fine, Mike. And remember that you started this! $\endgroup$ – Transistor Apr 7 '20 at 19:00
  • $\begingroup$ Just made the point so others check - I liked what you did and showed anyway so +1 $\endgroup$ – Solar Mike Apr 7 '20 at 19:18
  • $\begingroup$ Great, juste to be sure, the measure in your exemple would be 35.3mm, right? Does this method have a name? $\endgroup$ – Adrien Descamps Apr 7 '20 at 19:34
  • $\begingroup$ @Adrien: Yes. I was in a hurry and forgot the 1 mm numbers and addition. I've edited the drawing and chosen a different number with no repeating digits (35.7 mm). I don't know the name of the method. I was trying to think of one to see if the rulers were commercially available and couldn't find one in an image search. $\endgroup$ – Transistor Apr 7 '20 at 21:04

I would give a shot at image processing, if there is a good contrast between the surface you want to measure and the background. To do so you need to take a picture of the surface that includes an object of known size for calibration. Then, open source software like "imageJ" are able to perform the measurement. The accuracy of this method strongly depend on the possibility to have a photograph that is parallel to the surface to be measured. However, this can be partially circumvented by taking several photos, repeating the sizing, and considering the mean value.

  • $\begingroup$ This is a good idea, and probably the best method, but as you say, not easy to implement correctly. Making repeated measurement will probably be time consuming. I could of course calibrate the camera, and even automate the measurement with image processing, but i do not want to go that far for now. $\endgroup$ – Adrien Descamps Apr 7 '20 at 19:19
  • $\begingroup$ This is quite a complex solution and calibration would be required to account for lens distortion and object being farther away at the edges of the field of view. Industrial vision systems have to take this into account. $\endgroup$ – Transistor Apr 8 '20 at 16:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.