Whether license plate recognition, automated quality
control (Quality 4.0), monitoring production processes or inspecting
barcodes: Selecting the right camera sensor for industrial image
processing poses challenges for machine vision users. In addition to
considering various parameters such as sensor type, interfaces,
resolution, frame rate, exposure time, sensor size and quantum
efficiency , the exposure method (shutter) also plays an important role.
1. Sensor types and shutter
In its original function, the shutter
shields the camera film from light and opens when the shutter release
button is pressed. Depending on the exposure time set, the image then
appears brighter (longer exposure time) or darker (shorter exposure
time).
The same principle can
be found in the CMOS or CCD sensors that are commonly used today instead
of film. These read out images electronically by converting light
(photons) into electrical signals (electrons). Each image is composed of
a large number of horizontal lines, which in turn consist of a
different number of pixels depending on the resolution.
CCD
sensors are suitable for applications that do not require high frame
rates (max. 30 fps) or high resolutions (max. 2 MP) and achieve very
good image quality even in low light conditions. Due to their special
properties such as longer exposure time in dark skies, high measurement
accuracy or high data depth, CCD sensors are still of particular
interest in the scientific sector (e.g. astronomy, microscopy).
Sensors with CMOS technology, on the other hand, are
considered for applications that require high resolution and high frame
rates and are therefore particularly suitable for industrial image
processing as well as in the ITS area. Because of their low cost and
fast readout speed, CMOS sensors have become widely accepted over CCD
sensors. In addition, they convince with lower energy consumption, lower
heat production, improved quantum efficiency, improved noise properties
and a good price-performance ratio. The older CCD sensors are equipped
with global shutter due to their design. With the more modern CMOS
sensors, a distinction is made between rolling shutter and global
shutter sensors according to the type of exposure.
A rolling shutter sensor does not expose an image as a whole, but with a time delay according to rows or columns. This means that the image is not read out for all image pixels simultaneously, but sequential. With a global shutter sensor, on the other hand, all pixels, i.e. the entire image, are exposed at the same time.
2. Problem: Rolling shutter effect
The differences between global shutter and rolling shutter sensors are most apparent in dynamic image capture:
While
a global shutter sensor correctly displays moving images, since the
moving subject is photographed as a whole (snapshot of the entire
image), rolling shutter sensors can cause geometric distortions and
deformations (so-called rolling shutter effect).
This
occurs because the moving object is already at a different position on
the image sensor for each line during sequential readout. When all lines
are combined into one image, the moving subject then appears crooked or
distorted. In addition, when taking pictures under brightness
fluctuations (e.g., due to the flickering of artificial light sources
that is invisible to the human eye), dark and light stripes may appear
on the image.
It is true that
the rolling shutter effect can be partially mitigated by fast image
readout rates, elaborate flashand ambient light shading constructions,
or subsequent processing with appropriate programs.
However, the additional cost levels outweighs the basic price advantage of rolling versus global shutter sensors, and flash constructions only produce usable results in dark environments, but not in continuous lighting.
So
for tasks that require high image accuracy it makes sense to use global
shutter sensors from the outset to avoid poor results or subsequent
cost, hassle and time.
3. Solution in the industry: Global shutter
For this reason, it is imperative that CMOS sensors in the industrial image processing sector are equipped with a
global shutter. From our point of view, meaningful quality and
production controls using machine vision solutions are only conceivable
under this premise. Especially under industrial conditions (no daylight,
automated and fast motion sequences, vibrating and moving machines,
dust, etc.) and also in license plate recognition in the higher speed
range, rolling shutter cameras are very prone to errors and lead to
distorted or inaccurate images. The rolling shutter effect is therefore
particularly noticeable here, making image evaluation more difficult or
even impossible.
In contrast to many other manufacturers,
Wahtari offers the option of equipping all nCam variants with global
shutter sensors. This allows our nCam to produce razor-sharp images even
in high-frequency processes (e.g. assembly line production) or high
speeds (e.g. traffic monitoring), which can be easily evaluated by the integrated AI.
Wahtari
relies on the image sensors of the Alvium series from Allied Vision.
This includes numerous models with high-performance image sensors from
0.5 to 20.4 megapixel resolution and forms a solid foundation for the
highest image quality.
With
global shutter sensors from Allied Vision, Wahtari relies on strong
quality directly at the lowest level of image recognition to optimize
recognition accuracy in image evaluation while maintaining production
speed.
