The lighting system is an extremely important part of a machine vision system, and its quality directly affects subsequent image processing. Before attending a lecture by a Japanese lighting expert, I didn't really understand lighting. Wasn't it just about illuminating an image so the camera could capture it? But it's much more than that. Lighting is far more than simply increasing image brightness. A good lighting system can reduce a lot of image processing work and improve the efficiency of the entire machine vision system. So, what exactly is lighting? And how do you choose the right lighting system for a machine vision system?
Proper lighting is crucial for the success of machine vision applications and should be considered first. A well-designed lighting system not only delivers better performance and saves time but also reduces costs in the long run. Here are eight tips for choosing the most suitable machine vision lighting:
(1) Please use high-brightness light to detect material defects;
(2) For precise positioning, please use light of an appropriate wavelength;
(3) To inspect scratches on glass, please use non-diffused light;
(4) For testing transparent packaging, please use diffused light, i.e., diffused light;
(5) Use colored light to create contrast;
(6) Use a strobe light to detect fast-moving objects;
(7) Please use infrared light to eliminate reflections;
(8) To eliminate color changes, please use infrared light;
How does lighting affect machine vision applications?
Machine vision systems that prioritize output quality rely heavily on image quality. High-quality images enable the system to accurately interpret the information extracted from the detected object, resulting in reliable and repeatable system performance. The image quality required in any vision application is largely dependent on lighting conditions: color, angle, and the number of light sources used to illuminate the object mean the difference between good and bad images, potentially leading to better performance or poor quality images with undesirable results.
Machine vision lighting should maximize feature contrast while minimizing other remaining contrasts, thus allowing the camera to clearly see parts or markings. High-contrast features simplify integration and improve reliability; poor-contrast images and irregular lighting require more effort from the system and increase processing time. Optimal lighting depends on the size of the object being inspected, its surface features and partial geometry, and system requirements. A wide range of wavelengths (colors) and fields of view (sizes) allows for flexible selection of machine vision lighting to meet specific application needs.
The following five aspects need to be considered when choosing lighting:
1. Is the surface smooth or uneven?
2. Is the surface dull or shiny?
3. Is the object curved or flat?
4. What are the colors of the barcode or label?
5. Is it detecting moving objects or stationary objects?
Tip 1: Use a bright light to detect material defects
For example, verifying whether there are deficiencies in plastic casting.
Verifying material defects in plastic casting applications is important for ensuring a good sealing surface. When there are material defects, you have insufficient conditions (such as insufficient material being poured into the model).
Lighting technology: Bright field
Bright-field lighting techniques rely on surface texture and flat terrain. Light encountering a flat, reflective surface is reflected back to the camera, creating a bright area. Rough textures or surface defects scatter light away from the camera, creating dark areas.
Tip 2: Use the appropriate wavelength for precise component positioning
For example, inspecting flipped chips and verifying the correct component orientation in PCB assembly are common machine vision applications.
Lighting technology: Bright field
To verify assembly issues, the chip orientation was illuminated using blue light wavelengths. This illumination technique relies on wavelength and coaxial illumination geometry. The blue wavelength (460 nm) effectively distinguishes silver and copper surfaces: copper absorbs blue light, resulting in a dark field, while silver reflects blue light, resulting in a bright field. Coaxial illumination geometry eliminates erroneous reflections: unwanted glare, reflections, and dark spots.
Tip 3: Use non-diffuse light to detect cracks in glass.
For example, detecting cracks in glass containers.
Lighting technology: Dark field
In this application, dark-field lighting is used to create a bright feature of interest that is easily detectable against a dark background. Light passes directly through a transparent bottle in a dark-field area. Most light penetrating a transparent object goes undetected by the camera. However, if the material is irregular, such as with cracks, some light will highlight this irregularity. In particular, scratches create an internal void where light refracts and reflects, scattering at many angles, including returning to the camera. This light transforms the hard-to-detect scratches into a bright feature against a dark background.
Tip 4: Use diffused light to test transparent packaging
For example, verifying the contents of blister packaging.
Lighting technology: continuous diffusion
Continuous diffuse lighting does not emphasize surface texture and variations in elevation. It provides a very large fixed lighting angle, allowing light to reach the object from multiple angles, thus eliminating reflections and shadows typically produced by non-directional or single light sources.
Tip 5: Use color to create contrast
A useful method for creating a high-contrast image in machine vision applications is to illuminate the object with light of a specific wavelength (color). For monochrome cameras, the wavelength of light can make features appear brighter or darker than those in color. Using a color wheel as a reference, you can choose a light of the opposite color to darken a feature, or choose light of the same color to brighten it. For example:
1. If you want to darken a feature that is red, use green light;
2. Using green light can make green features appear brighter;
3. Remember the difference between the engravings on aluminum under red and blue light.
Tip 6: Use strobe lights for fast-moving objects
When a rapidly moving object produces a blurry image, a strobe light is needed. Strobe width = field of view ÷ pixels / speed of movement.
Tip 7: Use infrared light to eliminate reflections
Machine vision systems rely on grayscale conversion in digital images. In many vision applications, ambient light introduces unwanted bright reflections, making it difficult or impossible to detect features of interest. Infrared light can solve this problem.
Tip 8: Use infrared light to eliminate color differences
Infrared light can be used to eliminate grayscale differences between colored objects. Dark objects absorb infrared wavelengths, creating uniformity, while others appear as shadows. This lighting scheme is advantageous for detecting inconsistencies in color or shadow variations.
Choosing the right lighting scheme for your machine vision system requires consideration from multiple perspectives. Selecting the best options from these tips and combining them with the characteristics of your system will surely yield better results! Lighting is a vast subject that requires gradual learning.
