At the heart of any vision sensor is an imaging IC. This opto-electronic device can be thought of as an array of tiny light meters arranged in a grid, each one recording the amount of light reflected from a corresponding section of the target. When the output of each “light meter” or pixel is graphically displayed in a two-dimensional array, the result is a digital image. Portions of the target that reflect a lot of light show up as white areas in the image, while sections that scatter little light correspond to dark areas.
The fact that the entire process depends on reflected light means that lighting is one of the most critical variables in machine vision. It follows, then, that the simplest and most effective way to make a vision sensor “see” better is to select the correct lighting for the specific application.
When it comes to lighting, the name of the game is to bounce light off the target in a very controlled way. The placement and direction of the light relative to the vision sensor - and the target - is extremely important. Controlling how light strikes the target amounts to controlling the way light reflects off the target and into the vision sensor. This equates to controlling which portions of the image appear bright and which appear dark. The goal is to create a lighting scheme that will generate maximum optical contrast between a feature of interest and the background; the “good” condition appears bright and the “bad” condition appears dark.
The factory floor is nowhere near an ideal setting for vision sensing because completely controlling all light sources in the environment is simply not practical. Ambient light in particular can be a major problem, causing unwanted bright areas in the image. Another variable to consider is part-to-part fluctuation in color or texture because this also can change how light is reflected. For example, some targets may have oil on them, some may have been scuffed or sanded, and so on.
Art and science of lighting
Ideal lighting is that which creates the optimal optical contrast between the target (or product feature under observation) and its background. Keeping these ideal lighting conditions consistent is key. Lighting is one of the most important technical steps in an application, as it influences how the features under inspection appear to the vision sensor.
The correct lighting will highlight the features under inspection, disregard background objects, and overpower any ambient light in the mix. In contrast, incorrect lighting could make the inspection less robust or, even worse, impossible to perform altogether. Lighting technique is part art and part science. The following tips address the latter part, offering guidelines for lighting selection and usage in various applications.
Step-by-step lighting selection
Achieving the correct lighting scheme for a vision application is worth the investment in time and testing, and it can be done in a step-by-step process.
-
The first step is to define the feature you wish to examine on your target object. Features are generally separated into four categories: color, shape, surface texture, and translucency. Once you determine which of these optical properties differentiates a quality part from a defective one, you can begin to determine the lighting technique that will best solve your application.
-
The next step is to choose the illumination source best suited for detecting these qualities, usually from three lighting technology options: LED, halogen, or fluorescent. LEDs are generally selected for illuminating small objects at close range, but in some cases they are a good choice for long distance applications. LED-based lights are favored for their long life and compact size. Halogen light sources are also sometimes used, as they can deliver ultra high-intensity illumination in small areas and can provide structured lighting for objects.
Fluorescent lighting is used to create large, evenly illuminated areas. One note of caution: When using fluorescent lighting, take care to use a high-frequency ballast. Fluorescent fixtures used in a home or office environment typically turn off and on 60 times per second. While this on-off pattern is too fast for the human eye to notice, a vision sensor with a short exposure time can detect it, making some images appear dark (when the fluorescent lights are off) and some appear bright. A high-frequency ballast, in contrast, turns off and on thousands of times per second, making it a better fit for use in machine vision.
-
Another critical step is to select the correct lighting geometry for your application. Lighting geometry refers to the physical relationship between the light source, target object, and vision sensor. The angle of your lighting can be manipulated to produce varying light reflections, thereby improving the quality of the image your vision sensor observes. While many options exist, a great number of applications can be solved by one or more of the following lighting devices: ring lights, area (or directional) lights, low-angle lights, backlights, highly-diffused (or dome) lights, and on-axis lights.
Lighting solutions to reflect on
For many applications, the easiest lighting solution to implement — the ring light — offers sufficient illumination to create the optical contrast required. Ring lights mount directly to the vision sensor, with the sensor's lens positioned in the center of the ring. Ring lights illuminate the area directly in front of the sensor and are an exceptional solution for inspecting small objects and detecting color variations.
Area (or directional) lights mount separately from the vision sensor, and are placed at an angle to the target object, providing even illumination on a concentrated area. Area lights are often used to create shadows to highlight changes in a part's height, making them ideal for surface texture inspections. Area lights can also provide sufficient illumination on objects sitting more than a foot away from the lighting source.
Low-angle lights are also employed in surface texture detection applications, placed nearly perpendicular to the surface of interest and in very close proximity. Low-angle lights, when used with very small areas, enhance the contrast of surface features and emphasize changes in elevation, highlighting shape differences and surface irregularities such as dust, dents, and scratches.
On-axis lights are placed between the vision sensor and the target object, and a beam splitter directs the light rays along the same axis in which the sensor is looking. These lights deliver even, diffused illumination and are also good at detecting differences in surface texture. The image produced by an on-axis light is the inverse of that produced by a low-angle light.
Backlights provide even, low-intensity illumination and create the highest contrast of any lighting technique, making them the most accurate alternative. Backlights are placed behind the target object and aimed directly toward the vision sensor. These lights are best used on a part that can be separated from its background, and they are designed to highlight differences in translucency or to verify an object's proper size or shape.
Highly diffused (or dome) lights are aimed nearly perpendicular to the surface of interest and provide soft illumination from multiple directions. These lights are primarily used to minimize glare and shadows, uniformly illuminate curved surfaces, and minimize optical contrast due to variations in surface texture.
Several other considerations should be taken into account when selecting your lighting source, including the energy produced by the light, its durability and usable life, and whether continuous or strobed operation is best suited for your application. For more information on these options and to implement a solution in your plant, it is best to consult a machine vision expert. The most important thing to keep in mind is that finding the correct lighting for your application can be simplified by consulting these guidelines, but it will always be somewhat of a trial and error process — one that is critical to ensure your vision application's success.
For more information, contact Banner Engineering at (800) 809-7043 or visit bannerengineering.com. To read more about machine vision systems, how they work, and how to select critical components, visit motionsystemdesign.com's Knowledge FAQtory and look for links that will connect you to related articles and information.