The requirements when developing a machine vision system for line scan applications are diverse: conveyer movement, lighting, camera choice, lens choice and any image processing requirements must be harmonised for an optimal interaction of the system’s parts. Although lens choice is usually discussed during the system design process, in many cases the lens is dismissed as a minor component only needed to display the image on the camera’s sensor. Actually, it is much more than that. But how can the effectiveness of a lens be stated?

One of the best indicators for the effectiveness of a lens is the Modulation Transfer Function (MTF). The MTF number describes how well an image of fine black and white lines will be mapped with a lens. MTF measures how well contrast will be maintained as the thickness of the lines and the distance between the lines in the image gets smaller. The line density is denoted as lines per mm (lp/mm) and contrast decreases as the lines per mm increase until only a uniform grey is measured. This technique is comparable to a visual test at an eye specialist – a person with good eyesight will be able to read small letters, while a person with poor sight will only perceive small letters as grey spots.

The following drawings illustrate an MTF Measurement:

The design and the quality of the parts used in a lens are the determinants of MTF. The quality of the different vitreous bodies in the lens looms especially large. But what are the limits for the lenses in the machine vision world and what is the impact of pixel size in this context?

C-Mount lenses with a price between € 130 and 180 typically have an MTF of around 30 lp/mm and offer only a low contrast. 30 lp/mm is equivalent to a resolution of 17 μm. With these lenses, a change from black to white will be mapped over two or three pixels when the pixel size is, for example, 6.5 μm. In this case, it doesn’t make sense to use a camera with a smaller pixel size because object edges would only be mapped across more pixels.

Better quality lenses for machine vision sell for around € 400–700 and offer an MTF of about 70 lp/mm. This is equivalent to a 7 μm pixel size and represents rather good contrast. Good lenses have up to a 100 lp/mm MTF, the equivalent of a 5 μm pixel size. These lenses offer the perfect resolution for cameras with 5 μm pixels like the Basler L400k and L800k line scan cameras.

It’s true that smaller pixels can’t collect as many photons as larger pixels. But when you talk about line scan cameras such as the L400k and the L800k, you must also consider the short exposure times normally used in line scan applications.During these short exposure times, it is not possible to collect many photons regardless of pixel size. Therefore, the combination of 5 μm pixels and a high quality lens offers the ideal blend of performance indicators for most line scan applications. Also, smaller pixels result in a smaller camera housing size and this makes it much easier to cascade the cameras.

If you compare two line scan sensors with the same length but with different pixel sizes, you will get more image dots per unit length from the sensor with smaller pixels. When the lens is optimised for the system, the sensor with the smaller pixel size will provide more image information in comparison to the sensor with bigger pixels.

All of these considerations show that the choice of a lens for your application represents much more than simply displaying an image on a camera’s sensor. Because every application is more or less unique, it is impossible to provide specific recommendations that are true for all applications. But it is clear that to achieve the best performance from your system, you should carefully harmonise all the components you select for inclusion.

Christoph Noth Vision Components Product Manager Basler AG Tel. +49-(0)4102-463 0 christoph.noth@baslerweb.com www.baslerweb.com