Hybrid cameras – the future of image processing?
August 03, 2012
The miniaturization of products and huge cycle times in automated production are pushing conventional measuring instruments
and empirical data capture to their limits. However, in terms of speed in production and traceability of the data obtained,
image-processing solutions – in other words, optical measuring systems – hold the greatest potential. At the same time,
manufacturers are faced by major challenges when it comes to data capture and further processing, since the question arises:
at what point in image acquisition does image-processing offer a technically useful option and to what extent is hybrid generation
a sensible solution?
For smartphones to do what they do, there’s often more technology housed in just one centimeter of phone than was crammed
into an entire cell phone some years ago. That has been the case for a long time now and not only for smartphones;
the level of technical complexity of products that surround us every day is being constantly raised. As a result,
intelligent and accurate measuring and testing options are in demand for product development, production and quality assurance.
For high-speed camera technology, this means that the route from image to creation of the result, i.e. the conclusive analysis
obtained from the image acquired, has to allow for both the speeds and the accuracy that prevail in automation.
With today’s structures – whether it’s from the interfaces, via the connection to FPGAs, processors and memories – the search
is on for a feasible concept of image data processing for tomorrow. This is the challenge also facing Optronis GmbH, the German
manufacturer of high speed cameras for industrial application, where the main topic of discussion is hybrid generation: allowing
for the existing architecture of systems, at what point can we integrate data calculation via additional technologies?
Intelligent sensor systems
Basically, cameras differ in terms of specialty capabilities, i.e. image resolution and speed or intelligence.
Now, if the creation of the result is combined with the sensor, where the sensor is configured for the calculation
of the image processing, as a result compromises in capability are to be expected: a sensor converts the image into
electrical signals as it impacts on the sensor through the lens. The charge produced is digitalized via an analog/digital
transformer and using the image processor is merged into one digital image. Depending on the complexity of the calculation,
the processor does of course require capacities for processing. These work against the speed and image resolution factors.
Hybrid generation using sensor technology has already been successfully realized in a number of camera models and is an
attractive alternative mainly for large quantities and in terms of price/performance.
Frame grabber and PC
In the high-end segment of machine vision, where image detail, realtime and flexibility are called for,
the obvious approach is to swap out the creation of the result to a connected PC. The data is fed to the
PC via a frame grabber that is able to take over the precalculations. Normally digital frame grabbers are
equipped with programmable functions. There are also fully programmable systems based on FPGAs.
Therefore, the preprocessing of the data will have already taken place in the upstream frame grabber,
such as adjustments in light intensity and contrast, scaling and also complex image calculations.
The creation of the result itself is ultimately carried out in the PC’s processor. The advantage here is that today’s
interfaces in the PC (for example, PCI Express) are very powerful and the creation of the result can therefore take place
at the full readout speed of the camera. Of course this is of particular importance in the highly automated production
environment and with the extremely fast processes that need to be recorded in realtime. Hybrid generation would therefore
take place even outside the camera, between the frame grabber and the PC.
The decision starts with the problem
“Requirements and applications are extremely customized, particularly in the industrial environment.
That’s why it’s important as a manufacturer supplying modular and standardized products to be able to
offer an optimum solution. Of course, that requires the fundamental compatibility and standardization
of the platforms,‘ says Bernd Reinke, Industrial Cameras Division Manager at Optronis.
Therefore, according to Optronis, it is not the decision to adopt a specific technology that counts but
the flexibility to be able to respond to requirements – both technological and product-related.
“In just the same way as there might not be a perfect solution to interface discussions such as CoaXPress,
CameraLink or USB, you need to keep a constant an eye on the challenge the application presents in this
situation as well,‘ emphasizes Reinke.
“Optronis supplies industry in the high-speed sector. That’s why we are committed to CoaXPress
and modularity of the products. Other technologies might offer greater advantages for other sectors.
You simply have to get yourself some good advice and then decide which solution provides more benefits.‘
More information at: