Lighting & Viewing Methods
for Industrial Machine Vision
Options & Choices
Machine Vision
owes its versatility in large part to the variety of different lighting-viewing
methods (LVMs) that are available. One of the
distinctive features of Machine Vision is that we are very often free to employ
whatever LVM best suits the application requirements. Within the synthetic
environment of a factory, there are often several options for the
lighting-viewing conditions available, thereby allowing us to optimise image
quality. We can, for example, adjust
á The
number and spatial distribution of the light sources placed around the widget.
á The
spatial pattern of light projected onto the widget.
á The
intensity of each light source,
á The
spectral emission of each light source. (While we can control the colour of the
incident light, we are not limited to using only the VIS wave-band.)
á The
polarisation of the light impinging on the widget,
á The
spectral response of the camera, by placing suitable optical filters in its
optical path.
á The
polarisation of the light entering the camera.
á The
temporal pattern of light projected onto the widget,
á The
motion of the widget relative to the lighting-optical system and the
lighting-optical system relative to the camera..
á The
spatial coherence of the light projected onto the widget.
Optical Devices
We can also
employ a wide range of optical devices to enhance the image:
á Mirrors.
These may be plane, convex, concave, or anamorphic.
á Lenses.
These may be convex-convex, convex-concave, plano-convex, plano-concave,
Fresnel, or cylindrical. They may be used singly, or in tandem, Micro-lens
arrays are also available.
á Polarisers.
These may be linear, circular or elliptical.
á Prisms.
These come in a variety of shapes providing a range of functions.
á Filters.
á Beam-splitters
and devices made of birefringent materials.
á Gratings
and diffraction filters
á Quarter-wave
plates
á Diffusers,
These may rely on transmission or reflection...
á Fibre-optics.
These may be flexible, or rigid, often in the form of a rod or plate.
Types of Cameras
Several different
types of image sensor can be used
á array-scan
á line-scan
á circular
scan
á random
(program controlled) scan
á laser
scanner
á range
sensor.
An image sensor may
detect radiation in any of the
following spectral bands
á Visible
wave-band (monochrome)
á Visible
wave-band (colour)
á Multi-spectral
(
á Infra
red (IR)
á Thermal
(part of the IR wave-band)
á UV
radiation
á X-ray
Alternatively, image
sensors may be based on detecting
á Eddy
currents
á Electron
beams
á Magnetic
field
á Microwaves
á Neutron
beams
á Nuclear
magnetic resonance
á Pressure
á Tactile
sensors
á Ultra-sound
Wide Range of Options
We can use the
natural motion of artefacts as they are being transported around the factory,
or arrange for the widgets to be moved in some controlled way past the camera. Such
a wide range of options provides considerable versatility to design of the
image acquisition sub-system, However, it does increase the difficulty that the
vision engineer faces in choosing the optimal lighting-viewing method (LVM).
Despite this, some general principles have been discerned and are represented
here as a series of distinct LVMs, which can be used singly, or in various
combinations.
Pay Careful Attention to the Lighting
The task of
finding the optimal LVM is often perplexing. New-comers to Machine Vision
frequently make the mistake of restricting their attention to familiar domestic/office
illumination equipment (e.g. a desk lamp), believing implicitly that machines
see things in the same way as people do. Most people grossly under-estimate the
wide temporal and spatial variations in brightness and colour that occur in natural
and environmental lighting. It is unrealistic to believe that software can always
compensate for poor lighting. Adopting a casual approach to lighting will
almost always make the overall system much more complex and far less reliable
than it should be. It is only by understanding the general physical principles
involved in image formation and appreciating the relative merits of a wide
range of LVMs that an optimal system can be selected/designed.
Finding the Optimal LVM
No catalogue like this can provide a complete
substitute for experience and careful observation. The most powerful weapon in
the vision engineer's armoury is his/her own eyes. But they must be used
carefully, bearing in mind the warnings that we have described. By moving a
point source of light around the widget and observing its changing appearance,
it is possible to gain a great deal of insight about which LVMs might be
appropriate. (To be most effective, this should always be done in an otherwise
darkened room.) Of course, this experimental approach cannot help us to select
one of the more exotic lighting techniques but it is always well worthwhile looking at
the widget first. In a majority of cases, explorative visual examination does
help the designer a great deal. The merits of several of the LVMs described
below were first discovered in this way!
Comments About the Diagrams
Some of the principal LVMs are presented here, in
general non-quantitative form, without reference to physical scale, or their
cost. In some cases, proprietary lighting units are available to implement
these ideas and, where possible, we have included photographs and/or other
details of such equipment. Some companies have kindly allowed us to include
sample images to illustrate the use of their lighting-viewing products, by
demonstrating what kind of images can be expected. Such a mention in these
pages does not endorse these products, implying the author's preference over
those made by competitive organisations. Failure to mention a company's
products should not be inferred as implying that they are in any way inferior
to those mentioned here.
Combinations of LVMs
Optical components; such as lenses, mirrors, filters
prisms and, of course, the camera are the basic building blocks of the image
acquisition sub-system. They can be assembled in a very wide variety of ways.
These web pages illustrates some of the standard LVM arrangements that we can
build but, as a person gains more experience, the potential benefits of
previously untried configurations become apparent. The LVMs illustrated here
must be tailored individually, to suit the needs of each application. They can
be used singly, or in various combinations. For example, it is possible to
perform a useful inspection function using back illumination at one wavelength
and front illumination at another. Alternatively, we might combine range maps
and colour images,, or x-rays and fluorescence. Clearly, we cannot list all
possible combinations and, in practice, a great deal of experimentation and
cunning is required, to obtain the best results. The entries in this catalogue
should be regarded as being merely prompts guiding the practitioner towards a
few possible solutions for a given application. Having thus identified a set of
plausible suggestions, he/she can then concentrate on evaluating them
experimentally, to find the best one.
Catalogue
á
The LVMs indexed below are presented in alphabetical
order, since it is not possible to define any usful
á
The reader is urged to become familiar with the
concept and use of Illumination Diagrams, as these are used extensively
throughout this catalogue. (See Chapter ???)
á Light
sources are represented graphically, using icons suggesting the use of filament
lamps. It should be understood, however, that in many cases, other light sources,
such as discharge tubes, compact fluorescent lamps and LEDs might be more
appropriate.
Click on the Links Below
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Incomplete
- no link (Method 127) |
|
|
|
|
Incomplete
- no link (Method 128) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Optical
Fourier Transform - no link (Method 111) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|