SPIE Membership Get updates from SPIE Newsroom
  • Newsroom Home
  • Astronomy
  • Biomedical Optics & Medical Imaging
  • Defense & Security
  • Electronic Imaging & Signal Processing
  • Illumination & Displays
  • Lasers & Sources
  • Micro/Nano Lithography
  • Nanotechnology
  • Optical Design & Engineering
  • Optoelectronics & Communications
  • Remote Sensing
  • Sensing & Measurement
  • Solar & Alternative Energy
  • Sign up for Newsroom E-Alerts
  • Information for:
SPIE Photonics West 2018 | Call for Papers




Print PageEmail PageView PDF

Electronic Imaging & Signal Processing

In the noise

Noise-based ISO gives measure of camera performance in low-light, high-noise conditions.

From oemagazine January 2003
31 January 2003, SPIE Newsroom. DOI: 10.1117/2.5200301.0007

International Standards Organization (ISO) sensitivity standards as applied to digital image sensors can be classified as either saturation-based ISO (base ISO; see oemagazine, January 2002, page 56) or noise-based ISO. Base ISO is a common measure of sensitivity for photon shot-noise-limited applications like studio photography. Noise-based ISO provides a valuable tool for describing the sensitivity of imaging systems when lighting might be insufficient or sensor noise is high, or both, as in the case of many entry-level consumer digital still cameras that incorporate CMOS sensors and no flash system.1

Noise-based ISO is captured by calculating the average signal level of the gray patch, subtracting the known offset, and dividing by the standard deviation in the patch.

In general, an ISO measurement gives the exposure index at which certain image-quality parameters are achieved. The exposure index EI is related to the focal plane exposure H(lux * s)by


In the case of noise-based ISO, the image quality parameter of interest is the signal-to-noise ratio (SNR)—probably the most powerful measure of the quality of a monochrome image.2 According to the ISO standard, an SNR of 10 gives an "acceptable" image and 42 an "excellent" image.

For the purposes of this measurement, it is useful to describe the SNR as


where L is the scene luminance in cd/m2, g is the incremental system gain in digital numbers per cd/m2 of luminance, and σ is the standard deviation of the image output from a 64 x 64 patch, given in digital numbers. For a color imager, the highest-speed color channel is normally used for this calculation, although the ISO standard describes a more complicated color-channel weighting scheme. The gain g near L is calculated using three measurements, averaging the slopes immediately above and below L:


where gn is the incremental gain near Ln and S(Ln) is the signal output in digital numbers from the camera corresponding to a scene luminance of Ln. However, if the camera being used gives raw pixel data and the electronic offset is known, a more traditional SNR measurement can be made, and equations (2) and (3) are unnecessary. In either case, a gray 18%-reflectance patch is a suitable target for determining SNR.

The noise-based ISO is given by


where f/# is the effective f-number of the taking lens, t is the sensor integration time, and LSNR is the scene luminance that gives the required SNR. (For a discussion of the constant, see last year's article.)

In a typical test scenario, we generate a data curve by stepping the lens f/# while keeping the scene luminance and integration time fixed (see figure). We adjust the gain of each image to give the same average signal level (this step would not be necessary if the luminance was stepped instead of the f/#).

When you stroll into your local camera store to buy your next digital, be sure to ask for a noise-based ISO series. If you are on a budget within an order of magnitude of mine, you probably won't get one. But if you could, you'd really know what you were paying for. Happy shooting!


1. International Organization for Standardization standard 12232, Photography—Electronic Still Picture Cameras—Determination of ISO Speed (www.iso.ch).

2. J. Janesick, Scientific Charge-Coupled Devices, SPIE Press, Bellingham, WA (2001).

Gloria Putman
Gloria Putnam is an applications engineer at Eastman Kodak Co., Rochester, NY. Putnam teaches SPIE short course #SC494-- How to Select the Right Image Sensor for Your Application.