Camera projection errors measuring length

[Insidium]

So here's the situation, say i record, presuming the camera is perfectly vertical, an object falling next to a ruler. I then find how many pixels that ruler is and from there i can translate how many pixels the object has moved into how far the object has moved. The question is how reliable is this method? I presume the length in pixels of the ruler would change if i put it in different position.

 

[CWatters]

Most rulers have a scale on them. So you should be able to calculate a reasonable relationship between pixels and distance.

Lets say the ruler is 1 meter long and in one photo the rule appears 1000 pixels long. Then each pixel is 1mm. If the object moves 137 pixels then it's moved roughly 137mm.

If the ruler appears to be say 957 pixels long then each pixel is 1000/957mm long. If the object has moved say 357 pixels it has moved roughly 357 * 1000/957 = 373mm

Obviously if the ruler and object aren't in the same plane there might be parallax errors that reduce accuracy.

http://en.wikipedia.org/wiki/Parallax

Overall the accuracy depends on how well you set up the experiment. Perhaps keep the camera some distance away and use a zoom lens.

Run trial experiments, eg so you can adjust the zoom to make the experimental range fill the image as much as possible.

 

[voko]

In no case should one count pixels in this setup. All lenses - especially those not built specifically for orthophotography - will have image distortion, progressively bigger toward the edges. Always make sure that the ruler is included in the photo, that's the only way to measure anything reliably. Even then you will have some perspective aberrations, but those can be minimized.