Essential characteristics of noise

Sunday May 10 2009

Sensor Insight
Introduction | Noise in shadows | Noise in mid-tones | Noise in highlights | Summary | Annex

The gray level output is often 0 or close to 0 for very low illumination, but even if the sensor does not receive any light, there still can be fluctuations in the measured signal (due mainly to electronic and thermal factors) that can produce a high noise standard deviation. Therefore the SNR in dB can be negative for low sensor exposure. This noise in the absence of light is called dark noise, which is directly related to the value of dynamic range. Indeed, dynamic range is the ratio between the illuminance attaining the sensor saturation and the illuminance for which SNR equals 1. This value is usually attained for very low illuminance. However, a higher dark noise can considerably change this value. The larger the amount of dark noise, the less reliable the signal in shadows, thus reducing the usable dynamic range of the sensor. Denote the value of the output signal standard deviation in absence of light.

The SNR in the shadows for a gray level can be approximated by

Note that this is a raw value. In addition, the noise in shadows is always amplified by the camera processing for rendering colors and tones (gamma curve).

Noise in shadows: even with no light the image is noisy, and has a poor SNR in shadows