新闻
LED lighting for machine vision: geometries, techniques, and configurations
Lighting is the component that determines whether a machine vision system succeeds or fails. Even before the optics and the camera, it is the lighting that establishes which features of an object will be revealed and which will remain invisible to the vision system. Choosing the right geometry, wavelength, position, and operating mode means delivering a high-contrast, repeatable, easy-to-analyze image to the software; getting these parameters wrong forces costly and often unreliable software workarounds.
This guide brings together the practical criteria a system integrator or industrial engineer can use to set up the lighting for a machine vision application. We start with the available LED lighting geometries and their typical use cases, then cover lighting techniques (backlight/front light, bright field/dark field), the choice between white and monochromatic light, the difference between direct and diffuse lighting, continuous/boosted/strobe operating modes, and finally the recommended configuration for each type of surface. Throughout, we take a system-level approach, in which optics, lighting, camera, and software are designed as a single integrated bundle.
Questions this guide answers
- What are the main LED lighting geometries, and when should you use them?
- Backlight or front light? How should the light be positioned relative to the camera?
- Bright field or dark field: how do you choose based on the surface?
- White or monochromatic light: which delivers more contrast?
- Direct or diffuse lighting: how do they affect shadows and reflections?
- Continuous, boosted, or strobe: which operating mode should you choose?
- Which applications are these techniques used in?
LED lighting geometries
Geometry describes the shape of the source and the direction from which light reaches the sample. Each geometry is designed to bring out a different type of feature: contours, glossy surfaces, engravings, raised edges. The table below summarizes the most common geometries among LED illuminators for machine vision and their typical use.
| Geometry | Functional description | Typical use case |
| Ring light | Ring of LEDs around the optical axis, direct or diffuse light | General inspection, OCR, flat objects |
| Dome (diffusing dome) | Diffuse light reflected from the inner surface of the dome | Curved, reflective, metallic surfaces |
| Flat dome | Like the dome, but with a flat profile | Same advantage as the dome, where vertical space is limited |
| Tunnel | Diffuse lighting reflected by the tunnel along its axis | Long, narrow objects: mechanical components, cable harness sections |
| Coaxial | Light directed onto a beam splitter and deflected coaxially with the optical axis | Flat, glossy/mirror-like surfaces, wafers, laser engravings |
| Bar light | Linear array of LEDs, flexible side mounting | Angled lighting (dark/bright field) |
| Backlight | Rear illumination for silhouettes | Dimensional measurement, presence/absence, transparent objects |
| Line light | Long, narrow beam | Line-scan applications, continuous webs |
| Spot | Light beam | Front lighting from a single direction |
| Telecentric collimated | Beam of parallel, non-diverging rays | Precision metrology, edge detection |
Telecentric collimated illuminators are a distinctive part of Opto Engineering’s lighting range. Pairing a parallel-ray collimated illuminator with telecentric optics creates a decisive combination for accurate dimensional measurement: the collimated beam eliminates reflections and stray light at the edges, producing sharp contours that can be measured down to the micron. Opto Engineering’s established expertise enables the synergistic design of telecentric optics and lighting, optimizing how the two elements interact and improving application performance.
Backlight or front light: how to position the light
The first lighting decision is where to place the light relative to the part and the camera. With a backlight, the light and camera are on opposite sides of the sample, which, for opaque objects, appears as a very high-contrast silhouette. With a front light, the light and camera are on the same side of the sample, and the surface facing the vision system is typically what gets illuminated.
| Criterion | Backlight | Front light |
| Positioning | Light and camera on opposite sides of the part | Light and camera on the same side of the sample |
| Feature located | Contour/silhouette (or interior of a transparent part) | On the surface facing the camera |
| Part surface | Opaque or semi-transparent | Opaque or reflective, any finish |
| Achievable contrast | High contrast | Depends on the finish, typically lower |
| Accuracy | Can reach the micron range (bright field) | Not suited to sub-pixel measurement |
| Main risk | Loss of information on the top surface | Hotspots, saturation, and shadows on glossy surfaces |
| Typical task | Dimensional measurement, edge/profile/shape inspection | Surface inspection, code/text reading, color |
The rule of thumb is simple: if you only need to check shape and size, a bright field backlight is the best choice; if you need to see something on the surface (text, defects, color), you need a front light.
Bright field and dark field: making use of the angle
Note a point that is often misunderstood: the lighting technique is not defined by the sample geometry alone. The contrast of features in the final image depends on the interaction between the lighting and viewing geometry, the properties of the sample, and the imaging optics. The same relative positioning of components can produce bright field or dark field depending on the optics and the sample.
In bright field, light from direct (or specular) reflections off the sample surface is captured by the optics and contributes to the image: smooth surfaces reflect toward the camera and appear bright, while the sample features you want to reveal (defects, codes, impurities, etc.) deflect the light and show up as dark areas. In dark field, light from direct (or specular) reflections off the sample surface is not captured by the optics and does not contribute to the image: only edges, scratches, and engravings deflect light toward the lens and stand out brightly against a dark background.
| Criterion | Bright field | Dark field |
| Part surface | Opaque/reflective/diffusing, or sufficiently rough | Reflective/specular (polished metal, glass, wafers) |
| Light angle | Low, close to the lens axis* | High; the light must clear the field of view* |
| Working distance | Standard, flexible mounting | Short, a few centimeters from the sample |
| Typical task | Flat printing, labels, OCR, surface checks | Scratches, engravings, edges, raised defects |
* This criterion generally holds in the most common configurations (entocentric or telecentric optics and flat sample surfaces perpendicular to the optical axis). There are, however, exceptions.
The same logic also applies to backlighting: bright field backlight is used for silhouettes, edge detection, and dimensioning, while dark field backlight reveals cracks, bubbles, and inclusions inside transparent materials, acting as a defect detection tool rather than a measurement tool.
White or monochromatic light: choosing the wavelength
Depending on the type of inspection and the color of the surface, it is essential to know how to choose between white and monochromatic lighting.
The guiding principle is the color contrast rule: monochromatic light of the same color as a feature makes it brighter (highlights it), while the opposite color makes it appear dark (suppresses it). This is how, for example, red light increases the contrast of dark lettering on a red can.
| Criterion | White light | Monochromatic light |
| Color contrast rule | Not applicable, preserves all colors | Similar color → feature highlighted; opposite → suppressed |
| Matching camera | Color (preserves color information) | Monochrome, lower-cost, and higher-resolution images |
| Chromatic aberrations | Present (different wavelengths) | Absent |
| Sensitivity to ambient light | High, hard to filter out | Low, narrow bandpass filters eliminate it |
| Typical task | Multicolored parts, color sorting, color verification | Maximum contrast on a feature, defect/code reading |
In short: white light is the right choice when the actual color is information worth preserving (color sorting, color verification, general imaging); monochromatic light wins when you need maximum detail, robustness to ambient light, and precision, at the cost of losing color information.
Direct or diffuse lighting: controlling shadows and reflections
Direct light comes from a point source with parallel or slightly diverging rays: it produces sharp shadows, pronounced specular reflections, and high edge contrast, but risks hotspots on glossy surfaces. Diffuse light is obtained through a diffuser or a dome: it creates soft shadows, reduced reflections, and uniform illumination, at the cost of lower contrast and lower light intensity (for the same electrical power).
Another effective technical solution for reducing reflections is the use of polarizing filters, which selectively eliminate specular reflections while preserving the diffuse component. Polarization is particularly effective for inspecting glossy, metallic, painted, or wet surfaces.
Continuous, boosted, or strobe: operating modes
Depending on line speed, acquisition times, and the features to be inspected, LED illuminators can be used in three operating modes:
- Continuous: the light is always on; simple to manage, with lower peak intensity and higher thermal stress, which accelerates LED degradation over time.
- Boosted: current is driven above the rated value for extended periods to increase light intensity. A high on/off time ratio can shorten LED service life.
- Strobe: a short pulse synchronized with the camera exposure, which pushes peak intensity far higher than in continuous operation, “freezes” motion at high line speeds, and reduces average thermal stress.
The selection criterion is practical. In high-speed applications, strobe mode is the preferred solution. For static applications or with moving objects, continuous lighting is sufficient. When greater light intensity is needed without adding the complexity of synchronization, boosted mode is an option, provided its impact on the lighting system’s service life is carefully assessed. Strobe mode is effectively managed by LTDV strobe controllers, which allow precise adjustment of peak current, pulse duration, and synchronization with the camera exposure.
Applications
The same techniques are applied differently depending on the industry.
Designing the lighting for a machine vision system means choosing geometry, position (angle and distance), wavelength, and operating mode based on the features to be detected and the sample surface. None of these parameters should be chosen in isolation: the light works together with the optics and the camera, and the final result depends on the integration of the whole system. This is the approach behind the range of LED illuminators for machine vision, designed to be paired with optics and cameras in a single, coherent solution.