Random error in readings comes from several independent sources: shot noise from the discrete arrival of photons or charge carriers (its variance equals the mean signal, so it is physically irreducible and dominates bright signals), thermal (Johnson) noise in the sensor and its resistors, read noise in the amplifier chain, quantization in the ADC, supply ripple, and coupled interference from nearby switching circuits. Each source calls for a different fix: longer integration or averaging for shot noise, cooling or lower bandwidth for thermal noise, filtering, decoupling, shielding, or better layout for the electrical sources. Fixed-pattern noise is not random at all and is removed by calibration frames. Distinguishing random noise from systematic bias decides whether averaging or recalibration is the remedy.
Digital image sensors are characterized by resolution (the number of pixels captured), noise (unwanted signal modification arising at capture, conversion, transmission, processing, or storage stages,…