Image sensors

Integration techniques increase sensor range

A key problem challenging designers of image sensors is imaging the extreme changes of light contrasts within a single field of view. While conventional CCD sensors cannot normally handle high dynamic ranges, specialized CMOS devices can be used, but at the expense of higher fixed-pattern noise and temperature drift. However, recent developments in image sensors and technology are increasing the dynamic range of such devices by using dual integration methods.
July 1, 2000
3 min read

A key problem challenging designers of image sensors is imaging the extreme changes of light contrasts within a single field of view. While conventional CCD sensors cannot normally handle high dynamic ranges, specialized CMOS devices can be used, but at the expense of higher fixed-pattern noise and temperature drift. However, recent developments in image sensors and technology are increasing the dynamic range of such devices by using dual integration methods.

By combining two scenes of moderate dynamic range, images from the LARS II sensor from Silicon Vision clearly show both dim and bright areas. The sharp image of the bulb filament shows the blooming resistance of the pixels.

Click here to enlarge image

"A common approach to extend the illumination range exploits the logarithmic voltage-current response of diodes or MOSFETs," says Tarek Lulé of Silicon Vision (Siegen, Germany). "These logarithmic sensors read out the compressed voltage by simple source followers that allow random access. However, sensitivity and local contrast are poor while fixed pattern noise (FPN) is temperature-sensitive and amplified exponentially," he says.

To overcome these disadvantages, Lulé and his colleagues have developed a locally adaptive image sensor that overcomes these disadvantages. As every pixel on the 368 x 256-pixel device contains an automatic shutter and two capacitors, the total dynamic range can be split into two signals, each with moderate dynamic range. "Because both signals are stored as in-pixel capacitances, correlated double sampling can be applied to reduce FPN and temperature drift," Lulé says.

The important feature of the imager is the ability of the pixels to adapt themselves to the local illumination conditions. Every pixel optimizes its integration time to make best use of the available voltage swing without saturating. Combined with low-cost analog components and signal processors, both images can be composed to yield a scene with high dynamic range.

Similar development announced

Interestingly, Panasonic (Secaucus, NJ) announced a similar development last year. Panasonic's Super Dynamic technology incorporates a double-speed CCD and a DSP, imaging a scene every 1/60 s. While the first image is exposed at 1/60 s, the second is exposed between 1/1000 and 1/4000 s. Just as the long exposure images darker portions of the image, the short exposure captures the brighter portions. The two exposures are then combined to form an image with a dynamic range 64 time greater than that of CCD-based cameras.

So that optimum gain can be applied to the long and short signals independently, Super Dynamic II technology incorporates two separate automatic gain circuits so that optimum gain can be applied to the long and short signals independently. According to the company, this technology is now used in the WV-CS854, a 1/4-in. CCD security camera that incorporates a 3.8- to 83-mm autofocus zoom lens and a pan/tilt mechanism.

Sign up for Vision Systems Design Newsletters

Voice Your Opinion!

To join the conversation, and become an exclusive member of Vision Systems Design, create an account today!