In last month’s column, I talked about the accuracy funnel, which shows that in a measurement system where the sensor or the part is moved in one or more axes, the greater the distance, the larger the potential inaccuracies in the measured distance. In the case of an imaging system as used in a video or vision measurement system, there are other accuracy considerations that are important even when there’s no motion.

What am I measuring?
To understand what can affect accuracy at this level one must understand that video systems do not measure parts—they measure images of parts.

Yes, you see the part in an eyepiece or on a monitor, but what you see is an image of the part. Fundamental? Maybe. But it is important to the understanding of sources of measurement error in imaging. The entire image area is called the field of view.

A video measuring system can measure distances and angles within its field of view. The size of the field of view corresponds to the magnification of the optical system. Because most camera detectors today are rectangular, the field of view is typically rectangular. There are a number of constraints about how much an image can be magnified and how fine a detail can be resolved in that image (resolution), but more about that later.

Optical aberrations
Any imaging system is made up of lenses. Remember the handheld magnifying glass, or those single-lens magnifiers used for visual inspection? They magnify an image, but the amount of magnification depends on both the part-to-lens distance and your eye-to-lens distance. Change either the part or viewing distance, and the magnification changes or the part goes out of focus. The important thing to remember about such a simple lens is that nothing is being measured. The objective (no pun intended) is simply to increase the image size.

Lenses used in imaging systems for video measurement are made up of numerous optical elements that perform as a group to magnify and relay the image of the part to the camera. You might assume that all lenses are the same. If only it were that simple!

Lens quality and accuracy
Everyone is familiar with a house of mirrors at a carnival, where mirrors distort reflections to exaggerate parts of an image for humorous effect. What is happening is that your eyes see the reflection as if they are in an image plane, but the distances across the mirrors vary, so reflections that are closer look larger; those that are farther look smaller. Lenses can do the same thing. Of course, they do not distort to the same degree, but the concept is the same. The image through the lens may not faithfully represent the item being imaged.

There are numerous optical aberrations—spherical and chromatic aberration, coma, and astigmatism, for example—and all lens systems have these aberrations to some degree. The higher the measurement accuracy of a system with a lens, the smaller these aberrations must be to avoid influencing measurements. In other words, an inexpensive lens designed for photography may not be good enough for high-accuracy dimensional metrology because of image distortion from optical aberrations.

How aberrations affect measurements
Remember the house of mirrors? That same condition where light reflected from a plane is not imaged in a plane can affect measurement accuracy. Lenses are curved to collect light and focus it. Doing this symmetrically across the entire field of view should result in an image where all the points are in a plane. However, any number of conditions can prevent that from happening. Individual lenses may not be totally symmetrical. Spacing between lenses may not be exact. Individual lenses may be tilted relative to the optical axis through the lens. Any or all of these conditions can exist in a lens assembly used for video metrology.

Images that do not focus in a plane appear distorted. Consider a common lens aberration, barrel distortion. This effect makes the rectangular shape appear bowed outward at the edges, but has minimal effect at the center of the optics. With such a lens, measured points near the edges of the field would appear to be farther apart. Pincushion distortion has the opposite effect. Edges appear to pull towards the center. Well-designed and -built systems minimize these aberrations so they have little or no effect on measurements.

Focus effects
So far I’ve talked about the image plane. Unfortunately, unless the parts you are measuring are flat, the features you need to measure don’t lie in a plane, and this can affect measurement accuracy too. Back to the basic single lens, remember that even slight changes in lens position along the optical axis cause the image to increase or decrease in size. Now consider imaging a part with indents and columns on it. It’s possible that the area between the indents and columns is in focus while the tops and bottoms of those features are not. A measurement of the width of one of those columns, for example, may not be accurate due to the slight defocus. This effect is exaggerated at high magnifications, where the depth of focus is narrow. A telecentric lens can eliminate this source of error.

A telecentric lens ensures that objects in the field appear at the correct size even when not exactly in focus. Of course this does not work over large distances, but the effect can be dramatic when compared to a nontelecentric lens at the same magnification. That an image that might appear to be in focus may have this variation in magnification is worth knowing when analyzing measurement accuracy across systems.

Zoom lens accuracy
Zoom lenses make measuring easier because features of many sizes can be measured on a system without the need to change lenses. Of course, to make accurate dimensional measurements with a zoom lens, you need to know exactly how large the field of view is every time you change the magnification. A system could simply multiply the apparent zoom position by the known magnification at some point within the zoom range, but is that accurate enough? Is the zoom position known accurately? Is it repeatable (for example, does the lens come to the same position when zooming up as when zooming down)? For highest accuracy, you could recalibrate by measuring a calibration standard after each magnification change, but the time that takes would offset the advantage of the zoom function. The best solution when measurement accuracy is important is an automatic, internal calibration function. Not all systems have this.

Remember that telecentric lens? Telecentric zoom lenses are not always telecentric throughout their zoom ranges. A zoom lens might be advertised as being telecentric, and it very well may be, but there may be more to ask. The advantages of telecentricity are worth enjoying throughout the zoom range—caveat emptor.

A lens is a lens, isn’t it?
Another thing to consider in this discussion of distortion and errors is that there can be variations from lens to lens for supposedly identical lens assemblies. This means that a measuring system optimized to work with a particular lens may not perform the same if that lens is replaced with another of the same model. This is especially true if the lenses have asymmetric distortions. To maintain the highest accuracy measurements, always recalibrate a system whenever the lens is changed.

As you can tell, there are quite a few things that can affect measurement accuracy of systems that use lenses and cameras. It is safe to say that not all measuring systems are created equally. The optics are worth looking into (sorry…).

Of course measurement matters.