The past few columns about inspection and measurement have led to discussion of different systems for measuring based on images. One type of measurement done via image analysis is dimensional, or coordinate, measurement. The systems are called video, or vision measuring, systems. Here is a basic description of what these systems are and how they are used.
Why video measurement?
Video measurement is noncontact. Only light touches the part. A magnified image is measured. Video measurement is good for big parts, too. Scales and encoders track the positions of measured points anywhere across a part, even when those points are measured at the highest magnifications. Video measurement can be fast. CNC-based systems with metrology software can measure widely spaced coordinates on a part in seconds. Video measurement is accurate. Micron-level accuracy and repeatability are increasingly common. Video measurement excels in measurements in a plane, and it can also measure in three dimensions. Video measurement is versatile. It can measure parts made of virtually any material. Video measurement lets manufacturers ensure that processes are under control and that parts are within specification.

The proper level
There are three levels of video measurement system functionality:
• Manual
• Semi-automatic
• Automatic

Manual measurement
Manual video measurement is a step up from magnified vision inspection where a part is viewed, but nothing is measured. Here the part is placed on a translation stage, the image is focused, the part is positioned relative to a reticle or some other reference, an indicator display of some kind is zeroed, the part is translated to another location, and the distance or angle relative to the starting point is read from the display. The image is magnified either by a fixed magnification lens, a zoom lens, or selectable lenses of differing magnifications, such as in a turret. The part is typically viewed from above and the image viewed on a monitor of some kind. A light source may help with image brightness.

Semi-automatic measurement
Semi-automatic measurement isn’t a common product category term, but it’s descriptive of the capabilities of some more advanced video measuring systems. If computer-based measuring software is used with a manually operated system, part measurements can be programmed so other operators can follow a particular sequence for the measurement. In addition, image analysis tools can identify edges automatically. Software can guide proper focusing. The part is still moved on a manually operated stage, but where and how far to move the stages can be program-driven. As you might expect, semi-automatic measurement relies less on the expertise of the system operator to achieve accurate measurements than the manual process, so measurements vary less from operator to operator.

Automatic measurement
Automatic video measurement systems step beyond semi-automatic measurement through use of motorized, CNC stages and motorized lens systems. The computer-based measurement software allows detailed measurement routines to be created, saved and repeated. Because operation is motorized, those measurement routines can operate without any user intervention—truly automatic. Productivity increases because the machine does the work, and the quality control process is more stable because measurements are performed in the same way for every part on the system tested with that measurement routine by any operator.

So what makes up a video measurement system?

Imaging
Because it’s the image of the part that’s measured, good quality optics must faithfully present the image to a camera. Aberrations in poor quality optics might be interpreted as a measurement error in the part. Zoom lenses allow rapid magnification changes to maximize detail of features of various sizes. For highest accuracy, each change in magnification must be calibrated—the best systems do this automatically.

For software to analyze an image, it must be converted to electrical signals. This is done by “reading” the signal levels from each of the pixels in a digital sensor (camera). Best practice is to magnify the image until the feature of interest covers as many pixels as possible so individual details can be determined unambiguously. In addition, good video measurement systems use subpixeling algorithms to extend measurement resolution beyond the camera’s pixel resolution.

Positioning
For measurements beyond the optical field of view, it’s necessary to know the position of the image and the part throughout the measurement range of the system. Consider Z motion. It’s necessary to move the camera/optical assembly so the feature of interest is within the optical depth of focus at the particular magnification being used (depth of focus typically decreases as the magnification increases). The entire camera/optical assembly must stay in alignment as it moves. In automatic systems, Z motion is by motorized linear slides, with scales or encoders to keep track of position. On a micro scale, auto-focus takes over to maximize the sharpness of the image. System software notes the final position of the focused image relative to the original datum, so measured points are known within Z space.

Illumination
To maximize image quality at the camera, proper illumination is necessary. Because measured features can be straight or curved, and on the perimeter or the surface of the part, different methods of illumination are necessary. Back lighting (profile illumination) is best for through-holes and the perimeter of the part. Oblique-angle top lighting is helpful for highlighting subtle surface features. For example, a ring-light made up of concentric rings of LEDs allows illumination at selectable angles of incidence as the different rings are illuminated. More rings mean more angles. Segmenting the rings allows the light to be directional. This is especially helpful for features that lie in a particular orientation. More segments allow a wider range of illumination angles for maximizing edge contrast of features lying in virtually any orientation. The best-designed ring-lights focus the light coincident with the system optics so angle and direction of the light can be changed without moving the light in the Z axis. In every case, light intensity must be adjustable to avoid overdriving pixels with too much light or having low-light levels resulting in a poor signal-to-noise ratio, and associated poorly resolved edges.

Motion
It’s rare for an entire part to fit within the optical field of view. This means the part must be moved under the lens until every feature to be measured is imaged. Because this is a measuring machine, all motion in the X-Y plane must be quantified. X-Y stage travels of video machines can extend beyond one to two meters. Parameters of importance in stage motion include straightness of travel, stage speed and positional resolution. Open-loop motion uses counts of a known increment and assumes a position is reached based on the accumulation of a given number of counts. Closed-loop motion adds a feedback device such as a linear scale to actually measure the stage position. Total system accuracy depends on the type of motion system and the quality of the stage, motors, bearings, drive electronics and scales.

Structure
Measurement accuracy depends on the structural integrity of the system. The three axes must be orthogonal (true 90 degrees apart). This requires a precision design and exacting assembly. In addition, no part of the machine must move independently of any other or that offset will affect measurements. This requires a stable, damped mechanical structure, so video measuring systems are made of materials such as cast iron, steel and granite for the necessary structural stability.

System
Automatic video measurement requires choreographed motion of X-Y-Z stages; magnification changes; setting illumination type, intensity, and angle; and the acquisition and processing of camera data, which ultimately lead to a digitized model that contains dimensional and angular relationships of the part under test. There’s a lot going on, but user interaction with an automatic machine need not be complex.

What this means
Video measurement isn’t as widely understood as vision inspection. Inspecting a part to see if the label is missing or the bottle is full is easily understood. Using oblique LED illumination to highlight a faint edge that crosses an injection molded plastic part, so its relationship to a mounting hole tens of millimeters away can be measured with a few microns accuracy, is more complex. However, manufacturers in nearly any industry that need to measure or monitor dimensions, coordinates and angles use video measurement systems for just such things every day.

Next time, what about video measurement accuracy?