What an image processing board (IPB) is and what it does:
An image processing board – sometimes also called an image engine – uses digital signal processors, FPGA programmable logic and dedicated imaging ICs to obtain video streams from image sources such as a CIS, a CCD or a CMOS camera to then further process the video streams, to enhance, identify and compress them, and finally to send the results to a host system.
Why image processing boards are superior to traditional image acquisition:
A traditional image acquisition system consists of a digital image source, an ordinary frame grabber device installed in a powerful PC workstation and image processing and recognition PC software. An additional FPGA- or GPU-based hardware accelerator card may improve overall system performance beyond mere PC performance. At BAPIS, we have been developing a different image processing concept for over 20 years. Instead of relying on the PC’s processing power, we’ve been developing embedded, specialized image processing boards, combining the functionalities of frame grabbers, PC software and hardware accelerators – all in one device.
Where our image processing boards come into play:
BAPIS image processing boards find application in high-speed paper document archiving and recognition systems, in feature-rich desktop scanners, in automated election systems, in print inspection setups and in many more areas.
Image processing boards enable faster image processing, eliminate the need for high-performance PCs and can be integrated into almost any IT environment. They support the most widely used interface standards and reduce both the price and energy consumption of the overall solution. They are durable, long-lasting and built to work 24/7. In short, they are superior to traditional image capturing systems in every respect.
Immediate decisions: an embedded IPB allows immediate decisions to be made at the scanned object level. An example of such an immediate decision is content-based sorting, otherwise impossible or requiring a much longer paper path to accommodate a PC architecture and operating system lags and delays in a traditional image acquisition system.
Functional in any IT environment: an IPB, delivering complete results from its real-time deterministic firmware, allows the use of any mini-computer or low-cost PC as a host. The list of supported operating systems includes the most recent generations of MS Windows and a wide range of Linux distributions (also ARM-based), allowing the integration of our IPBs in various IT environments.
USB standard and various video source interfaces: in contrast to traditional frame grabbers, closely coupled with a specific PC architecture, our IPBs use a widely adopted USB standard interface for the delivery of the processing/recognition results. A superspeed mode and a high-speed mode are available depending on customer requirements. Our IPBs support various video source interfaces in order to accommodate customer requirements for speed, transmission distance and overall solution price. Camera Link (base, medium, full), G-Link & CoaXPress are among the already available source interface standards at BAPIS.
Dozenfold power efficiency: an IPB, designed and built for a specialized task, has much higher power efficiency than any traditional acquisition system based on a general-purpose PC workstation. Our IPBs outperform traditional PCs dozens of times in terms of number of processed images per single watt of consumed energy. That means distinctively lower operating costs and a concern for our planet‘s resources.
Durable, long lasting, functional 24/7: the IPBs without any moving or easily breakable parts are durable, long-lasting devices, built to work 24 hours a day. The absolute majority of all the image processing boards manufactured at BAPIS over the past 20 years are still in operation in our customers‘ systems.
Long manufacturing life cycles: we care about the long-term availability of all our products – IPBs being no exception. The manufacturing life cycles of our IPBs reach up to 12 years, depending on the model.