To offer the broadest range of products, high-speed camera manufacturers must continually trade off factors such as resolution, sensitivity, and speed. To increase sensitivity, for example, imagers with 12-μm pixels or larger are often used, which effectively decreases the achievable resolution of cameras based around them (see “Capturing the Moment,” Vision Systems Design, April 2009).
As well as performing a tradeoff between these two variables, camera vendors are also considering adding smart features in the form of automatic shutters, smart lenses, and high-speed interfaces that will make their cameras easier for the system developer to deploy.
One vendor, Vision Research (Wayne, NJ, USA; www.visionresearch.com), recently unveiled plans for its next generation of camera systems that will employ many of these techniques. “In ballistics monitoring applications, high-speed cameras may be located at long distances from the host computer in environmentally challenging locations,” says Rick Robinson of Vision Research.
In these cases, dark current calibration can be challenging. Previously, this dark current calibration was performed manually by placing a lens cap over the camera lens, capturing an image, and using it as a reference from which subsequent images were subtracted. “Since this dark current is proportional to temperature,” says Robinson, “the thermal dark current will double with approximately every 6° increase in temperature. In desert conditions where multiple cameras may be deployed, this may mean manually calibrating each camera every hour to compensate for thermal dark current.”
Engineers at Vision Research plan to incorporate an electromechanical shutter into their next generation of cameras in response. “In this way,” says Robinson, “the operator will be able to automatically calibrate the camera for thermal dark current effects by programming the camera over a Gigabit Ethernet interface.”
To simplify aperture control and focusing of its next generation of cameras, the company also plans to incorporate a method of automatically controlling these functions. Canon (Tokyo, Japan; www.canon.com) has already developed an interface standard that allows Canon camera bodies (such as the 16.7-Mpixel EOS 1Ds Mark II) to electronically control functions such as autofocus and aperture with its electronic-focus (E-F) lenses (see “Smart components add functionality,” Vision Systems Design, September 2005).
Third-party companies such as Birger Engineering (Boston, MA, USA; www.birger.com) have successfully reverse-engineered this interface in products such as the EF232 adapter that allows Canon E-F-mount lenses to be used without requiring a Canon EOS camera body. Vision Research also plans to incorporate this interface to allow Canon’s lenses to be used with its range of high-speed cameras.
“Although these Canon lenses have primarily been developed for 35-mm backplanes or digital sensors with 22 × 16-mm formats,” says Robinson, “they will be able to be used with cameras such as Vision Research’s Phantom v310 that uses a 25.6 × 16-mm imager.”
As well as offering a standard EOS interface, Vision Research is also developing an external controller that will allow its current range of cameras to be retrofitted with the interface. Of course, one of the most important aspects of any high-speed camera design is the maximum frame rate and the maximum number of frames that can be stored at any given speed. At present, many camera manufacturers use on-board memory to store image sequences before they are transferred over slower interfaces such as Gigabit Ethernet to the host computer.
Like many other camera companies, Vision Research is also planning to incorporate a 10 GigE interface into its cameras, although, admits Robinson, this will still not be fast enough to transfer 1280 × 800-pixel images at 3000 frames/s from its v310 camera to the host computer. To achieve this data rate, Vision Research is also considering developing a standard interface that supports data rates of at least four times that of the current Camera Link standard (see “Camera Link 2 awaiting market drivers,” page 15). Like BitFlow’s proposal, Vision Research’s implementation likely as not will allow a number of pre-existing serial protocols such as InfiniBand or Serial RapidIO to be used.
