Essential characteristics of noise

Sunday May 10 2009

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

At higher light levels, another phenomenon appears. Each pixel has its own analog gain factor, which is directly related to the physical size of certain capacitors. This phenomenon is called Pixel Response Non-Uniformity (PRNU).

The resulting noise can be modeled as additive, with a standard deviation proportional to the signal, i.e., . The SNR in the highlight is approximately

where is a constant which is characteristic of pixel non-uniformity.

This implies that the SNR does not increase further when PRNU becomes the main source of noise (which could always happen because PRNU increases faster than photonic noise). However, sensors always have a saturation value and PRNU is usually not the dominant source of noise at this value. Moreover, PRNU is a fix bias and can be compensated for with a calibration.

Color sensitivity is an integral value over the whole curve characteristic. In particular it depends on the noise in highlights. However, it also depends on the sensor’s spectral sensitivities which determines how the sensor can see the colors of the scene.