Camera Interface Standards: New Choices, Next Directions
January 2003,  by Barry Mazor

Here's a free truism for the New Year: If you don't interface the front end of your imaging system for maximum effectiveness, you're going to feel it throughout system performance. Since this will not be news-but remains a fact of life that affects so many imaging system designs calls-the state of camera interfacing and what you can expect next in that arena is our first Roundtable topic for 2003. We've found the answers provocative, and trust that you will, too.

Where the growth is
Advanced Imaging: No matter which interfaces your firm may be employing-USB 2.0, Firewire, Camera Link-which of these do you see as providing the best opportunity for growth in 2003-and why? Plus, are there specific camera markets or combinations of markets driving that growth?

Steve Nordhauser, Silicon Imaging: We see significant growth opportunities in high-speed and megapixel digital vision cameras in markets where traditional CCD and video format solutions fall short. Our USB 2.0 cameras provide an easy solution to replace the traditional analog cameras in machine vision and microscopy systems. With Camera Link, we can achieve extreme speeds at 12-bit precision for applications requiring both high resolution and high speed, or simply for live video rates, such as film scanning, biometrics, PCB inspections and tracking.

Mike Miethig, DALSA: There will be growth for all of these interfaces. But Camera Link was designed by a team of camera manufacturers and frame grabber companies to meet the specific needs of the machine vision market. It supports all types of cameras-line scan, time delay and integration, area scan, color or monochrome-to any resolution with data rates from a few megabytes per second to 640 megabytes per second. Customers who require higher resolution cameras with greater bit depths and ever-faster data rates are ideally positioned to take full advantage of Camera Link's capabilities.
Vladimir Tucakov, Point Grey: There have been very few UB 2.0 cameras available in 2002, so the growth of this particular interface, technically speaking, will be the greatest. However, Firewire will still provide the best opportunity for growth as the interfaces are cheap, the technology is mature and robust. Given the uncertainties in the economy, high-end Camera Link solutions may sell, but they also may not, if companies do not increase their spending on capital assets.

Fabio Perelli, Matrox Imaging: Currently, we do not see much support for USB 2.0 in industrial applications. But we have seen significant growth in the use of both Firewire and Camera Link standards, and we are forecasting further growth for these in 2003. The cost of Firewire cameras has decreased, making that standard a suitable choice for lower cost or standard-performance applications, such as those which require lower frame rates or lower image resolution. The Camera Link standard delivers better performance for higher-end vision applications.

Don W. Lake, PULNiX: Camera Link is by far the largest vehicle for growth where performance, measured in pixels per second, is the driving parameter-as it is for the ever-higher spatial resolution cameras and the ever-higher frame rate cameras, and especially when the two are combined. Firewire, as always, remains the great potential interface and cannot be ignored. USB 2.0 would seem to be more significant in late 2003 and beyond if only because of a lack of available product today. Both Firewire and USB 2.0 target low-cost markets and applications where performance need not be great or can be compromised.

Scientific/industrial impact
Advanced Imaging: For demanding scientific/industrial imaging apps, where are these three camera standards succeeding-and where do they fail, in your experience, to meet end user expectations?

Fabio Perelli, Matrox Imaging: Firewire has seen success in lower cost applications where lower image resolution and lower frame rates are acceptable. Alternatively, Camera Link is successful in more demanding applications, those with high frame rates or high resolution. In terms of ease of connectivity, both Firewire and Camera Link finally solve the cabling issue, facilitating the connection between cameras and frame grabbers. The one drawback that we still have, for both of these standards, is cable length.

Don W. Lake, PULNiX: Both Firewire and USB 2.0 are perceived to fail when bandwidth limitations impose the barrier to achieving a successful system. Camera Link usually provides enough bandwidth to shift the barrier to success away from the interface to other systems elements.

Vladimir Tucakov, Point Grey: Camera Link has succeeded at providing solutions for high-speed/high-resolution applications. Firewire has been successful at multi-camera applications. In particular, the ability of some of the cameras to automatically synchronize to each other has been useful for many 3D applications. It is not clear where USB 2.0 cameras will make a contribution at scientific and industrial applications.

Steve Nordhauser, Silicon Imaging: The Camera Link interface has significantly simplified the integration of cameras and frame grabbers, giving imaging component buyers the ability to chose the best of class for the camera and then the most suitable frame grabber based on speed (32/33 vs. 64/66), form-factor (PCI, CPCI, PMC PC-104+), OS-Support (Windows, Linux, etc.), high-level software application and cost. We see the USB 2.0 interface succeeding in applications where low-cost, small-size, laptop compatibility and minimum power consumption are the driving factors.

Mike Miethig, DALSA: CameraLink's advantage lies in its real-time triggering and ability to transfer data at very fast rates today. Camera Link customers benefit from the wide variety of camera products and frame grabbers-and the knowledge that Camera Link is an industrial standard with a long lifeline.

Next capabilities
Advanced Imaging: What should we all be watching for next from these three standards in working capability-for the likes of remote access, mobility across different computing platforms, smaller footprints, etc..?

Steve Nordhauser, Silicon Imaging: More intelligence will be put into the cameras to simplify frame grabber or host interface communication and control software. New mobile products will emerge using laptops and hand-held PDAs for capture, processing and display, all operating on battery power. And we see an increase in demand for non-Windows-based (i.e., Linux) embedded solutions for both Camera Link and USB 2.0-based solutions, for use in low-cost standalone inspection systems, portable image capture and high-speed recording and networking systems.

Mike Miethig, DALSA: Right now, the Camera Link committee is working to improve the software interface to the camera serial communications port, and introducing a time multiplexing definition that will double the data rate passed through the cable. Additionally, the committee is working toward self-certification to give customers some assurances of camera/frame grabber compatibility, and has discussed auto-configure capability.

Vladimir Tucakov, Point Grey: The size and cost of Firewire cameras will continue to shrink. USB 2.0 will certainly become the standard consumer interface, but it is not clear if it will be suitable for industrial applications. USB 2.0 and Camera Link would certainly benefit from a standard similar to the IIDC digital camera standard. Regardless, the days of analog cameras are numbered!

Fabio Perelli, Matrox Imaging: Since cameras cannot always be placed next to the PC, we believe that increasing the cable length is a necessity for many industrial applications. That needs to happen next.

Don W. Lake, PULNiX: Remember-these standards do not drive needs; they make the solutions to those needs practical for everyone. Whatever the end-user needs today will appear within the standards tomorrow. The crucial point, then, is to make sure that the appropriate standards bodies and the committees within them are sensitized to emerging needs as quickly as possible.

About that PC attachment...
Advanced Imaging: Do you foresee a continuing trend toward more direct connectivity between cameras (and other peripheral devices) simplified via cheaper silicon solutions? Or do you see this as more a matter of implementing software for all applications?

Steve Nordhauser, Silicon Imaging: No! We see the imaging solutions with our cameras continuing to be PC-centric. The PC or other computing device will continue to improve in speed and performance, enabling the cameras to becoming simpler, smaller and lower cost image acquisition peripherals. Only the complete stand-alone inspection camera sensor will incorporate these silicon solutions, but will likely be based on Ethernet connectivity.

Mike Miethig, DALSA: Camera Link will always need a frame grabber to receive and reformat video data into a format ready for processing. Data reformatting allows for smaller lower cost cameras and reduces CPU power requirements for processing. Other general purpose inputs and outputs necessary to an industrial vision application are provided by a frame grabber. Frame grabbers (and cameras) will continue to benefit from the ever-increasing capability of FPGAs at ever decreasing costs.

Don W. Lake, PULNiX: Cheaper silicon solutions will always be the preferred course; it's just not always available in a form applicable to imaging applications. Even then, engineers do really interesting things to make do with available cheap silicon in their designs knowing that however convoluted the hardware is, if it works and is less costly, it is often better than an elegant, expensive implementation.

Vladimir Tucakov, Point Grey: The peer-to-peer connectivity is certainly an important aspect of many advanced applications. FPGAs and DSPs provide great benefits to customers with requirements for large number of cameras where PCs may not be required or are not desirable.

Fabio Perelli, Matrox Imaging: USB 2.0 and Firewire ports are now standard on today's PCs. This adds tremendous value for both the software vendor and the customer, since the software communicates and/or controls the cameras. For the Camera Link standard, a software-only solution is not feasible; due to the complexity of Camera Link, especially with its support for multiple taps, a more "sophisticated" frame grabber is required to format the image data properly.

Supplying end-user software
Advanced Imaging: Some camera makers have been suggesting that it would be easier still to identify end-user software to supply with their products. What, as you see it, is the best approach to selecting app software to fulfill specific user criteria, such as usefulness for real time or turnkey applications?

Steve Nordhauser, Silicon Imaging: Most of our clients are OEMs who are developing software or systems for a targeted market; they want a simple development kit with an API to adjust the camera parameters and control the timing and capture of their images. These SDKs should have a limited up-front cost to support the initial development effort and then free distribution with the hardware. We would like our customers to integrate real-time or turnkey applications with our hardware as a component of the overall solution.

Mike Miethig, DALSA: At DALSA, we sell to many OEMs who view their value-add or expertise as the algorithms used to extract the image information. It is the small end user who is looking for a more complete solution; which may include the lighting, the camera, and the algorithms to solve his application problem. DALSA works with its network of value-add re-sellers to help the end-user find appropriate imaging solutions to their vision applications.

Vladimir Tucakov, Point Grey: The approach that our company has taken has been to develop drivers and SDKs that allow image acquisition and camera control with as much flexibility as possible. Being able to answer hardware, firmware, driver and SDK questions at a single point of support has been a winning combination for us. Having a camera that conforms to a digital camera specification has also been beneficial to providing a "plug-and-play" solution.

Fabio Perelli, Matrox Imaging: Many camera manufacturers have established relationships with frame grabber and software vendors. These relationships greatly facilitate the integration of the customer's vision system components. We believe that the greatest challenge today lies in selecting components that perform at the same level. For example, no software will increase a camera's frame rate, so it's important for the customer to know the application's requirements. As a frame grabber/software manufacturer, we provide application notes which describe how to interface different cameras with our equipment to obtain the desired performance.

Don W. Lake, PULNiX: Camera makers who pro-actively supply/recommend end-user software do so because their cameras integrate some special feature(s) that can facilitate the solution to a particular problem or class of problems. End-users with that problem then have only the task of evaluation of the camera makers' selected hardware and software. If it works for them, great! That's the best approach; if it doesn't, at least the investment to find out was minimal, so it was still the best approach!

top