Calibrate camera coordinates and colors

E-mail

Cameras used for metrology or color matching need calibration. Lenses can slightly distort images, and improperly mounting a camera can introduce measurement errors. Improper lighting or ambient light conditions can distort colors.

A machine-vision lens may exhibit barrel distortion, an effect that causes the sides of a square to bow out slightly in an image. This type of distortion can cause a 1–2% error between edges on an object and those in a camera’s image. If your application cannot accept a measurement error within that range, choose machine-vision software that will compensate for barrel distortion or buy a higher-quality lens that can reduce barrel distortion to about 0.1%. (When you talk with lens companies about lens distortion, you may hear a “promille” specification, which means parts per thousand. Thus, 0.1% becomes 1 promille.)

This Digital ColorChecker SG calibration chart provides 140 squares that represent a range of colors, gray scales, and skin tones. Although used primarily to calibrate photography cameras, the chart can calibrate color-vision equipment, too. Courtesy of X-Rite and GretagMacbeth.

A poorly mounted camera can produce perspective or trapezoidal distortion that makes distant objects in the camera’s depth of field look closer together. In metrology applications, you must calibrate for this distortion or mount a camera perpendicular to the surface of the object you want to inspect.

To determine the extent of distortion, you can use a standard chart that provides accurately spaced and sized targets. This information also lets you relate an x-mm distance to the number of pixels that represent that distance in an image. An Accu-Place Dot-Distortion Target from Applied Image, for example, gives a camera dimensional targets that range from 2-mm-diameter circles on a 4-mm grid in an outer zone to 0.20-mm dots on a 0.40-mm grid in the center. Engineers can obtain other types of dimensional-calibration targets, too.

You may not need to calibrate your vision system’s color measurements if you plan a simple comparison that distinguishes, say, white from red, because qualitative color information will suffice. But if your vision system must compare a color, perhaps on a keyboard or LCD, with a standard, you must calibrate the system’s color measurements.

You can buy several types of standard color charts, but the ColorChecker charts from GretagMacbeth provide one of the better ranges of colors in a pattern of squares. The supplier offers a corresponding list of RGB values (daylight illumination) for each square, so you can relate standard colors to the RGB values your camera produces and the RGB values that your vision-system software processes and displays.

jontitus@comcast.net