Machine vision The lighting system in the system is an extremely important part, and its quality directly affects the subsequent image processing. Before listening to a lecture by a light source expert, I actually didn't know much about lighting. Isn't it just illuminating the image so that the camera can capture the image? But this is not the case. Lighting is far from being as simple as enhancing 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 is lighting? How to choose a suitable lighting system in a machine vision system?

Appropriate lighting is the key to the success of machine vision applications, and it is the first part to consider. A well-designed lighting system will not only bring better performance and save time, but also save costs in the long run. Here are eight tips for choosing the most suitable machine vision lighting. They are:

(1) Please use high-brightness light for detecting material defects;

(2) Please use light of appropriate wavelength for precise positioning;

(3) To detect scratches on the glass, please use non-diffused light, namely Non-Diffused Light;

(4) For detecting transparent packaging, please use diffused light, namely Diffused Light;

(5) Please use color light to create contrast;

(6) Please use stroboscopic flash to detect fast moving objects;

(7) Please use infrared light when eliminating reflections;

(8) Please use infrared light to eliminate color changes;

How does lighting affect machine vision applications?

The machine vision system that will output the highest quality depends on the image quality. High-quality images enable the system to accurately interpret the information extracted from the detected object, so that reliable and repeatable system performance can be produced. The image quality required in any vision application largely depends on the lighting conditions: the color, angle, and the number of light sources used to illuminate the object means that the difference between good images may result in better performance and also Poor quality images produce bad results.

Machine vision lighting should maximize feature contrast while minimizing other remaining contrast, so that the camera can clearly see the part or mark. High-contrast features simplify integration and improve reliability; poor-contrast images and irregular lighting require more effort from the system and also increase processing time. The optimal illumination depends on the size of the detected object, its surface characteristics and partial geometric characteristics and system requirements. With a wide range of wavelength (color), field of view (size), for special application needs, you can flexibly choose machine vision lighting.

The following five aspects need to be considered when choosing lighting:

1. Is the surface smooth or rugged?

2. Is the surface dull or shiny?

3. Is the object curved or flat?

4. What is the color of the barcode or mark?

5. Is it to detect moving objects or stationary objects?

Tip 1: Use bright light to detect material defects

For example, whether it is insufficient to verify in plastic casting

It is important to verify material defects from plastic casting applications to ensure a good sealing surface. When there is a material defect, you have insufficient conditions (such as insufficient material poured into the model).

Lighting technology: bright field

Bright field lighting technology relies on surface texture and flat terrain. Light hitting a flat reflective surface reflects the light back to the camera, creating a bright area. Rough textures or surface defects will scatter light away from the camera, creating dark areas.

Tip 2: Use the right wavelength for precise component positioning

For example, detecting Flipped Chips and verifying proper part orientation in PCB assembly are common machine vision applications.

Lighting technology: bright field

In order to verify the assembly problem, use the blue wavelength to highlight the chip orientation. This lighting technique relies on wavelength and coaxial lighting geometry. The blue wavelength (460 nm) can distinguish silver and copper surfaces well: copper absorbs blue light and presents a dark field, while silver reflects blue light and presents a bright field. Coaxial lighting geometry eliminates false reflections: unwanted dazzling points, reflections and dark spots.

Tip 3: Use non-diffuse light to detect cracks in glass

Such as detecting cracks on glass containers

Lighting technology: dark field

In this application, darkfield illumination is used to create a bright feature of interest that is easy to detect against a dark background. In a dark field area, light passes directly through the transparent bottle. Most light penetrating transparent objects will not be detected by the camera. If the material is irregular, such as cracks, some light will highlight the irregularity. In particular, scratches create an internal void where light is refracted and reflected, scattered at many angles including returning to the camera. These lights transform the hard-to-detect scratches into bright features on a dark background.

Tip 4: Use diffuse light to detect transparent packaging

Such as verifying the content of blister packaging

Lighting technology: continuous diffusion

Continuous diffuse lighting technology does not emphasize changes in surface texture and elevation. It provides a very large fixed lighting angle, allowing light to find the object from multiple angles, thus eliminating reflections and typical non-directional or shadows caused by a single light source.

Tip 5: Use color to create contrast

A useful way to create a high-contrast image in machine vision applications is to illuminate objects with light of a special wavelength (color). For black and white cameras, the wavelength of light can brighten or darken the same features as color. Using the color wheel as a reference, choose a light of the opposite color to darken the feature; or choose a light of the same color to brighten the feature. E.g:

1. If the feature you want to darken is red, use green light;

2. Use green light to make green features appear brighter;

3. Remember the difference between red and blue light engraved on aluminum.

Tip 6: Use stroboscopic flash for fast-moving objects

When the image of fast moving objects is blurred, you need to use stroboscopic light. Strobe width = field of view ÷ pixel/moving speed

Tip 7: Use infrared light to eliminate reflections

Machine vision systems rely on the conversion of gray levels in digital images. In many vision applications, ambient light brings unwanted bright reflections, which makes 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 present shadows. This lighting scheme is useful for detecting inconsistencies in color or shadow changes.

Choosing a suitable lighting solution for your machine vision system requires consideration from many aspects. Choose the desirable one from these tips and combine it with your own system characteristics. I believe it will be twice the result with half the effort! Lighting knowledge is very large, and it needs to be learned step by step~!