Vision applications often involve performing both calibration and fixturing. You almost always calibrate an image so that you can report results in real-world values. You may also create a fixtured space that tracks object features so that you can make subsequent measurements of these features. If you need only an approximate calibration or if you know the exact dimensions of the part you intend to inspect, you can perform calibration and fixturing in either one or two steps. The following sections describe one- and two-step calibration and fixturing.

In one-step calibration and fixturing, you use the geometric fixturing method with an N-point fixturing tool to create a fixtured space in physical units. Unfixtured points are typically expressed in a pixel-based space and raw fixtured points are provided in physical units.

By default, the tool only includes rotation and translation in the calculations of fixtured space. To do one-step calibration and fixturing, however, you enable all degrees of freedom (DOFs) for the fixturing calculation. You then supply known real-world measurements for the object's features. Each time the N-point fixture tool runs, it recalculates fixtured space. But because you have enabled all DOFs, the tool considers scaling and skew in the calculation: it treats the known geometric measurements as equivalent to the scaling factor that would be computed by the first step of the calibration process.

One-step calibration and fixturing works for situations where you track and test for the presence or absence of object features. For example, you may simply want to fixture a ROI relative to the read-only tab of a 3.5-inch computer disk. You then run a Histogram tool to analyze the brightness levels of the region and determine whether the tab is in the read-only position or not. If the ROI changes in scale, this method keeps the fixtured space always relative to the feature's location.

Note, however, that you should not attempt to measure feature distances or sizes with the one-step calibration and fixturing method, unless you are absolutely certain about the physical position of the features used for fixturing. Suppose, for instance, that you accidentally acquire an image of a 5.25-inch floppy disk. Because you have enabled all DOFs (including scaling, aspect, and skew) and supplied object-local, physical coordinates for a 3.5-inch disk, the tool adjusts the scaling factor of the resulting fixture to make the 5.25-inch floppy be 3.5 fixture-units in size. This may still create a fixtured space that tracks a read-only tab, but you cannot validly measure the size of the 5.25-inch floppy. If you did, the floppy would appear to have the same dimensions as the 3.5-inch disk, which obviously does not correspond to the physical size of the floppy. It does not make sense, in this case, to try to measure the physical sizes and distances between object features that were initially used to define the meaning of physical space.

Although you rarely use one-step calibration and fixturing, you might choose it if:

• You have not already calibrated your input image.
• You know the physical dimensions of your object.
• You only need an approximate calibration

This last criteria may apply if you want to work in (approximate) physical units and:

• You are making only relative measurements (for example, measuring whether the disk hub is centered between the left and right edges of your floppy diskette).
• You are making non-geometric measurements (such as image brightness).

By using two-step calibration and fixturing, you can avoid the drawbacks of the one-step method. In two-step calibration and fixturing, you first perform calibration using the N-point calibration tool to map pixel coordinates to real-world units. You then feed the resulting output image, which has a newly attached calibrated coordinate space, to the N-point fixturing tool and use either reference fixturing or geometric fixturing to create a fixtured space. The N-point fixturing tool includes only rotation and translation in the calculation of the fixtured space. You have already supplied the scaling or skew factors from the calibration tool.

Choose two-step calibration and fixturing if:

• You need an accurate way to map image features to physical space.
• You have previously calibrated an input image.
• You intend to make measurements in physical space.
• You want to keep the calibration and fixturing operations separate, performing calibration only once at set-up time.