In order to achieve a cost-effective machine vision system, developers must choose the most suitable product from the many lenses available.

For machine vision system designers, because there are many lens products of different types, different resolutions and different image distortion characteristics available on the market, it becomes more complicated to make a cost-performance trade-off when selecting a lens for a particular application. Prime lenses are still the mainstay of many machine vision systems because of their low cost. There are many other lens options available on the market, including zoom, zoom, telecentric, 360° optics, and liquid lenses, each with its own unique advantages for specific applications.

Determine the focal length

Before choosing any particular lens, the focal length must be determined. The choice of focal length depends on the resolution required to image the defect, the size of the object being imaged, and the distance of the object from the camera. The focal length here refers to the distance between the optical center of the lens and the image sensor of the camera.

By changing the focal length of the lens, different sizes of the field of view (FOV) can be obtained. Choosing the correct focal length of a lens depends on the working distance of the object from the camera/lens system, the required field of view and the size of the image sensor. The focal length of a lens can be determined by the following formula:

Focal length = magnification × working distance/(1+ magnification)

Where magnification = sensor size /FOV.

Therefore, for the same working distance, the larger the sensor size, the larger the field of view produced. For example, the opal-2000 camera from Adimec of the Netherlands has a sensor size of 2/3 inch and an estimated focal length of 29.93mm for a 50mm (horizontal) FOV and 200mm working distance.

When selecting a lens, the resolution of the lens must match the characteristics of the camera's image sensor. Therefore, system developers must carefully match camera and lens.

Prime or zoom?

Prime focus lenses are widely used in machine vision systems because they use fewer optical components, have low optical distortion and are relatively cheap. However, in some applications, it may be necessary to change the field of view, especially if the system design may change over time, or if the system integrator needs to determine the appropriate focal length for an application. In this case, you can choose a zoom lens. Unlike zoom lenses, which keep the focus position the same when the focal length changes, zoom lenses need to be refocused while allowing different fields of view to be imaged. Zoom lenses are more expensive than prime lenses because they require a middle or rear assembly that slides back and forth, as well as a separate focus adjustment mechanism for the front unit. In the article Selecting Lenses for a Vision System, Lumenera, a Canadian camera supplier, states that it is prudent to select cameras for use in systems that have already been selected in advance: Consider buying a cheap (low quality) zoom lens to determine the right focal length.

Unlike zoom lenses, which need to be refocused, zoom lenses, or parfocal lenses, keep the focus position the same when the focal length changes. While many companies use the term zoom lens to describe their products, many of these products are actually zoom lenses that require manual refocus.

amplification

Although zoom lenses are not common in machine vision applications, such lenses are often used in applications such as microscopic imaging to provide manual or motor controlled image magnification. By using this zoom lens, the operator can check the part at the desired magnification without changing the lens; Or you don't need to have multiple types of lenses on a single disc (which allows you to change the magnification without loading and unloading the lens). With zoom lenses, the detection system can be automated so that the system can be programmed to view the entire scene at low magnification and zoom in on specific details without changing the lens or rotating the lens dial. The zoom lens can be adjusted either manually or automatically by computer control. Because such lenses change their focal length by moving the lens element, they have been shunned in the past by developers who require mechanical stability, accuracy and repeatability. Since launching these lenses in 2008, Excelitas Technologies has now improved the system design in its Fetura+ product with infinite correction for zoom, increased focus repeatability near the edge of the field of view, and improved optical performance, And compatible with the company's Optem Fusion lens system accessories.

Telecentric design

With traditional lenses, the closer the object is to the camera, the bigger the image looks. The farther away the object is from the camera, the smaller the image looks. This is a disadvantage in high precision measurement applications where the image processing software will measure the parameters of the part based on the captured images. To overcome this problem, system developers can use telecentric lenses in order to obtain images of objects of the same size, independent of the object's position in space.

The distance over which an object can move and still appear to be the same size after imaging is called the depth of field. Magnifying depth of field is different from image sharpness depth of field, which is commonly understood as depth of field.

Telecentric lenses are usually larger and more expensive than conventional lenses because they require more lens elements and the lens needs to be as large as the object being imaged. There are three types of telecentric lenses on the market -- object side telecentric lenses, image side telecentric lenses, and dual telecentric lenses.

Although many manufacturers offer these three types of telecentric lenses, image side telecentric lenses are more commonly used in image projection equipment and are less commonly used in the field of machine vision. In lithography systems, for example, projection lenses are typically image-square telecentric lenses used to image lithography masks onto silicon wafers. The advantage of such image-square telecentric lenses is that they provide uniform light transmission over the field of view.

In machine vision system, the most commonly used telecentric lens is object square telecentric lens and double telecentric lens. Object side telecentric lenses require fewer lens elements than dual telecentric lenses and are therefore less expensive.

The object side telecentric lens is telecentric on the object side; Bitelecentric lenses, on the other hand, are telecentric on both the object side and the imaging side, providing constant magnification even when the imager in the camera is not always guaranteed to be at the exact position in the optical path. This dual telecentric lens is often used in conjunction with a collimated back illuminator to ensure a high-contrast image, enabling accurate image measurements.

Lower camera costs

For applications such as beverage, pharmaceutical, and cosmetic testing, parts must be tested at high speed as they move along a conveyor belt at high speed. To do this, a system must be built that can image the sides of objects, the top of objects and even the inside of containers to detect defects. Of course, there are many different ways to accomplish this task.

The task can be accomplished by using multiple cameras to image the top and sides of the object. Here, multiple cameras, mounting brackets, and software calibration programs can be used. Alternatively, the object can be rotated around the field of view of a single linear or planar array camera to capture a 360° omnidirectional image of the object. This method requires more complex mechanical engineering, as objects must be stopped, rotated and tested before passing/failing.

To overcome this problem, you can use a PERICentric or Catadioptric lens. Unlike conventional lenses or telocentric lenses, the PERICentric lens is designed to allow light to converge towards the object being imaged. The use of such a lens in a machine vision system allows viewing of both the top and side of an object, thus reducing the number of cameras and lenses required, as well as eliminating the complexity of multi-camera calibration.

Electric focusing

While traditional lenses can be used to image objects at different distances from the camera, if objects of different heights appear in the imaging system, the lens will need to be refocused. While manual focus can be done on an automated production line, this requires resetting the camera for different focal lengths. Manual adjustment on the one hand will require adjustment time, resulting in increased downtime; On the other hand, if the camera and lens are located in inaccessible areas of the device, manual adjustment can be very difficult.

Although currently available lenses meet the needs of many machine vision applications, more specialized machine vision systems may require custom lenses and coatings. Fortunately, many lens manufacturers with in-house production capabilities are ready to customize lenses to meet these application needs. Of course, these custom lenses can be expensive, so they are usually used only in specific imaging systems that are not cost-sensitive (such as military applications) or in the production lines of mass consumer products.

Unmanned factories need machine vision