www.machinevision

.ie

English

Albanian

Arabic

Bulgarian

Catalan

Croatian

Czech

Danish

Dutch

Estonian

Filipino

Finnish

French

German

Greek

Hebrew

Hindi

Hungarian

Indonesian

Italian

Japanese

Korean

Latvian

Lithuanian

Maltese

Norwegian

Polish

Portuguese

Romanian

Russian

Serbian

Slovak

Slovenian

Spanish

Swedish

Thai

Turkish

Ukrainian

Vietnamese

Galician

machine vision illumination

machine vision illumination

Custom search

machine vision illumination

For anybody starting off in machine vision it quickly becomes obvious just how critical a role illumination plays in a vision solution. If you can light your target inspection product in the correct manner then identifying the unique features or faults you are looking for becomes a much easier task. If you can obtain a contrast level which makes obvious to the human eye the parts to be inspected then the confidence level of making a working product will be greatly increased.

ASPECTS TO CONSIDER IN MACHINE VISION ILLUMINATION

Cameras do not see objects. They merely see light which is reflected from objects. The intrinsic qualities of materials control how it reacts to different lighting. Therefore to get a lighting solution that would work reliably for the intrinsic properties of the material being inspected is the key. While this would be the primary consideration when considering a lighting solution, there are other factors which need to be investigated.

What speed will the application run.
What is the size of the physical area that needs to be illuminated.
What does the service life of the lighting need to be.
What are the mechanical constraints.
What are the environmental constraints.

It's critical that you choose the correct illumination for all 2D Vision applications. The rule of thumb is that if you can distinguish the features you want to extract from an image with the naked eye, then your software will be able to do it. You don't really need a full machine vision system to determine your chances of success. Just look at the captured image with the correct illumination and if you can easily pick out the features that you need to detect without angling it or magnifying it, then the chances are pretty good.

A badly illuminated image will make the software

Types of Illumination to consider

Bright Field

In bright field the light is shone directly at the object for inspection. If you need an image with good variation in contrast, bright field is the solution. It is not ideal when inspecting shiny surfaces. See co-axial lighting for highly reflective surfaces. The light is reflected into the camera. See the image below.

Dark Field

With this illumination approach light is shone at an angle to the surface which will pick up any variations on the surface of the item to be inspected. If the surface to be inspected is flat and no damage or dimples the light won't be reflected back to the camera resulting In a dark or constant image. Any small defects will show up as bright spots. As you can see from the image below, no light is reflected directly into the camera, so there are no 'Hot Spots'.

Back Lighting

In some applications for measurement or accurate defect detection and object orientation a silhouette of the object to be inspected is needed. It is very important to have a constant light distribution across the area to be inspected. The light is placed behind the object. A typical backlight is shown here.

Diffuse Light

Diffuse light or sometimes called cloudy day illumination is used when non- directional light is required I.e. no shadows. This light is used for objects with a shiny surface finish. The image below is aluminium and the objective when lighting it is to ensure that there is no reflection.

Co-Axial

This is similar to the light obtained from the cloudy light source. The light is shone at a beam splitter and the reflected light returns tothe camera perpendicular to the initial incident ray as can be seen in the diagram. This light is perfect for shiny surfaces and also areas with varying small shadows. A good example is the top of a drinks can which is an inverted dome. Reading the text on this curved surface would need a light source like this.

machine-vision-illuminationa

Click Image to enlarge.

How to find the right illumination

Illumination is probably the most crucial part of any machine vision application. If you get the lighting correct then the images you will be processing will be images with good contrast and be easy to analyze. Do this incorrectly and you will be left attempting to use software to correct for poor images and make it much harder to find a solution.

In general you will need to maximize the contrast of the features of interest. Minimize the contrast in features you are not interested in and minimize external influences. Lets look at the difference between human vision and machine vision. In the case of a human being they possess a complex vision system of their own with the ability to learn from experience. They have the ability to move a part around to obtain the best lighting even when lighting is at an inadequate level.

A Machine vision solution will generally only have static performance data. Predetermined positions for image acquisition and predetermined position of lighting, limited range of motion, predetermined functionality of the software. Given these limitations the machine vision system will find its result in a reliable and stable manner.

While there are items that can be seen by the human eye which cannot be seen by machine vision and vice versa observing the part with the human eye using different lighting techniques is a good starting point.

machine-vision-light

Click Image to enlarge.

A short video showing a DOAL light on a shiny surface

doal-light

Click Image to enlarge.