DIY Digital Caliper Micrometer Comparator

Intro: DIY Digital Caliper Micrometer Comparator

Get precise measurements with a cheap desktop scanner.

Using free software, you can measure part sizes with a resolution of less than a mil ( 1/1000 inch or about 25 microns ).

Great for identifying mystery micro-screws, electronics, camera parts, or even measuring plants or insects.

Tips and tricks for mounting parts, scanning, aligning the image, and taking measurements.

I'll give you links to other resources where people are using scanners for measurements and would like to hear from you about any ideas or feedback on how to get better results.

STEP 1: Free Software for Measurement

You can use nearly any graphics software for taking measurements.

The Notes and Resources step of this instructable gives links to notable image measuring software. Three free choices are IrfanView, The GIMP, and ImageJ. All are easy to use for basic measurements.

For simple measuring of lengths,  I prefer IrfanView graphics software.

Keep in mind that The GIMP is excellent and is my main image editor. It is worth your time to learn it as a free alternative to Photoshop.

ImageJ is a scientific and math miracle. Easy to use, but loaded with incredible features. Thanks to i'bler aman0311 for suggesting this.

IrfanView

IrfanView is Windows-only freeware. It is easy to load and use. It is kept up-to-date by its author Irfan Skiljan.

Be sure to download it from the official site IrfanView or from a trusted site like C|Net or tucows.

It installs quickly and easily. It's very light on resources, and is easy to learn.

STEP 2: The GIMP

The GIMP is another favorite Open Source Freeware graphics program useful for measuring.

It runs on Windows, Mac, or Linux.

It's very powerful, and for basic measuring is easy to use, though not as simple as IrfanView.

You can download the GNU Image Manipulation Program from the official GIMP site or from C|Net and other trusted sources.

You can also use Photoshop, Picasa and many other programs to directly measure scanned parts.

STEP 3: Testing Your Scanner

Every scanner is different, so it's important to do some basic checks to see how yours works.
One simple check  finds the center of the scanner sensor array.
Usually this is a path about an inch or more wide running down the center line of the glass.
Use paper tubes as shown, or LEGO bricks.
Place them on the glass with the scanner lid up.
Run a preview scan.
Now see the next step.

STEP 4: Calibration and Tilt

Look at the preview scan below.

The paper tubes on or near the center line show the best place to scan lumpy (3D) small objects.

The tubes further away all seem to lean in towards the center line. This is because the scanner sensor is mounted on the center line and uses mirrors and a lens to capture the entire width of the glass.

Note that there is no tilt up or down. Only sideways. Sort of like walking down the middle of a road and seeing the trees and buildings leaning in over you.

It also helps to place an accurate drafting ruler in different positions to check for accuracy.

STEP 5: How to Mount Lumpy Objects

Flat objects can just be placed on the scanner glass.

But a round or lumpy object like a small bolt or screw should be fixed in place.

I use a LEGO brick and some Blu-Tack poster mounting putty.

I place the LEGO brick on the center line of the glass at the bottom of the scanner glass.

This gives the best lighting angle with the fewest shadows.

You'll have to experiment because all scanners are different.

STEP 6: Cover With Paper and Scan

Use plain white paper over the glass as a backdrop.

You can try different colors depending on what you are scanning.

If you leave the lid open without any paper, you will get a deep black background which can be good for shiny parts.

STEP 7: Preview Scan

Set the scanner for photographs, color and 2,400 dots per inch.

After running a preview scan (see picture), choose a small region for the final scan.

STEP 8: Straighten the Image

Use the rotation tool in your graphics program to straighten out the image.

Most of the time you will find you need to rotate the image 2 or 3 degrees to make the part look straight.

For IrfanView, press F12 to bring up the paint tools window. Choose the straighten tool, trace a line on the part which should be vertical. This rotates the image.

In the GIMP, drag a guideline or turn on the grid. Now use the Rotation tool to free rotate the image.

STEP 9: Measure the Part

Once the image is straight, you can begin measuring.

In IrfanView's paint menu, choose the measurement tool.

Hold down the Ctrl key to snap the measuring line to horizontal, vertical, or 45 degrees.

You can now write down the measurement results or paste them into a spreadsheet or text file.

STEP 10: Notes and Resources

Notes
* HowStuffWorks How Scanners Work
* Wikipedia Desktop Scanner

Software
* IrfanView  View Convert Measure
* The GIMP GNU Image Manipulation Program
* USA National Institutes of Health ImageJ Java Image Processing
* VistaMetrix Image Measurement Software
* Planimeter Measure Image Areas
* Screen Calipers On-screen Measurement Tool
* Ruler for Windows Pixel Ruler

Comments

[Light_Lab]

I can recommend this technique, we have used it in the lab for about 20 years. The first time I used it was to measure thin flexible polymer sheet that was impossible to measure accurately any other way.
We never rely on firmware interpolated resolutions though, we always work at the fundamental mechanical resolution of the scanner. We also check the scanner by scanning an engineers steel rule. We have found the occasional scanner firmware interpolation that is inaccurate, and some cheaper scanners that have inaccurate DPI in their mechanics, sometimes they are non-linear.
It seems that the more expensive scanners are the most accurate and precise, we currently use an Epson V700 with a transparency back (highly recommended) to get 6400dpi, that is ~4 micron per pixel; so you can get ~25 pixels across the average width of a human hair.
Currently it is used to measure experimental soft contact lenses under development.

[petr77]

Would not taking a digital picture with say a smartfone (decent quality, rigged to a microscope perhaps) achieve the same or better results than with a scanner?

[Light_Lab]

Short answer: No, not for really high accuracy.
(It does depend on the size of what you are measuring and how accurate you need to be. I am assuming you mean for really accurately measuring parts and components and larger objects up to the size of the scanner.)
Even if errors introduced by the physical positioning of smartphone and target could be avoided; lenses without some degree of distortion don't exist; and lenses in phones are generally of poorer quality than those used in cameras or microscopes. The scanner avoids lens distortion problems by mechanically scanning the whole image step by step.
The other thing is the resolution: A really good quality camera (phone or otherwise) can provide 5000 pixels over the whole image. This means it can only compete with the scanner for measurement resolution up to about an inch. A typical camera macro field of view is ~2", so you definitely would have to add some sort of magnifying or microscope lens to the camera or phone to beat the scanner.
BTW Generally macro lenses introduce more distortion.
Obviously once you get down to really microscopic items, a microscope with a camera (and proper calibration) is going to be a better "Micrometer".
Just one last note: There are image processing techniques to almost totally remove lens distortion and you can also put a transparent grid plate over the item when photographing it to supply local calibration references. Nothing is totally impossible.

[Light_Lab]

Smart phones are not known for having good quality lenses. Even the best lenses in the most expensive cameras suffer from distortions that are not visually objectionable, but will have a deleterious effect on accuracy when measuring dimensions. Nevertheless if you keep an accurate scale such as an engineers ruler very near what you are measuring you may get sufficient accuracy.

Good quality scanners have very accurate linear drives that make them far more accurate than any image obtained through a lens.

[aman0311]

flattness is an issue, but can be measured by acquiring an image of a known square object, and then drawing a straight measurement line from one sode of the object to another. Bottom line is this boils down to the quality of the optics used in acquisition. Shooting from farther away past a certain point (usually the lahf-way travel through a typical zoom lens) won't improve things. Basically take any lens assembly, set it at it's half-power point, and back up until you can focus in the midrange of the focusing system. Once there your in the best spot to acquire.

[robotmastern]

 very nice. Ive been needing a caliper. this will work in a pinch but i really need to go to the science and surplus store and get a real one

[kelseymh]

What a really fascinating application of image software!  And for people who already own an "all-in-one" printer, much cheaper than buying precision machine tools :-)  Well documented as well.

[iectyx3c]

Thanks kelseymyh. Probably everyone already knows how to do this. I should try googling for photogrammetry and scanner to see what others have done.  I imagine there are better ways than my cheap and cheerful tip.

[kelseymh]

I doubt it.  If you're in a heay technical field where metrology is crucial (building silicon-based particle physics detectors, for example), then maybe it's obvious. 

If you asked the 20,000+ members of I'bles about it, I'd say there might be a few hundred who knew about the concept, and maybe a couple of dozen who could figure out how to make it work.

Your I'ble is a great introduction to a very sophisticated subject.

[isacco]

Congratulations for this Instructable! It is useful and well explained.
The idea is simple and smart. It is not a real novelty, but the scanner calibration is a good idea for non flat objects.
I am in the research field and scanning flat objects (examples, leaves or plant roots) for automatic measurement is a usual task. Specialized software are available for this, both licensed and freeware.
A very good free software for complex measurement on digital images is ImageTool
ddsdx.uthscsa.edu/dig/itdesc.html
Bye
:)
Isacco

[aman0311]

You can also use the NIH freeware that runs on Java called ImageJ. We use this frequently in microscopy.

Image J can perform calibrated measurements both hand-drawn as well as morphometric measurements (like mean feret, max/min, area, stuff like that). 

Finally the technique you used to calibrate can also be used with any digital camera. Simply take a picture of a known standard, find the pixel calibration, don't move the subject to camera distance and you can smap + measure away. This is best done using an improvised copy stand.

[iectyx3c]

Thanks aman0311 for the very useful ideas. Thanks to you, I am exploring the simplicity and power of ImageJ. I have added it to my list of favorite measurement software with a link to you for credit. ImageJ is an entire scientific toolbox of learning and is very impressive. I am grateful to you for telling me about it. I also like your camera calibration method and will try it out. I imagine that zooming in from a long distance probably gives the flattest image for measuring?

[mbudde]

That seems to be very useful. And it's FREE! What could be better?

[iectyx3c]

I think lots of people must already do this. I just find it to be a quick way to measure parts. If you guys find any i'bles or web sites already about this, please post them here and I will link to them. Thanks.

[Ninzerbean]

I have no need for this but it was a really fascinating 'ible to read, thank you for sharing your knowledge. It's funny but the title is what made me what to know more.

[iectyx3c]

Thanks for your comment - yeah, I love browsing and reading i'bles in areas totally foreign to me, like sewing, cooking, or whatever, because I get ideas for my projects. i'bles is a great creativity booster.

[Trebawa]

Idea: if the object is static, scan it a few times and combine them in the GIMP to eliminate the effects of noise and hopefully improve the (effective) resolution.

[iectyx3c]

Smart idea. Thanks for suggesting. Definitely would work.