CMOS Image Sensor Noise Analysis Through Noise Power Spectral Density Including Undersampling Effect Due to Readout Sequence

This paper presents a noise analysis in the frequency domain using noise power spectral density and considering the undersampling process coming from the native readout circuit sequence of image sensors. It reveals aliasing effect impacting the noise level of the image sensor. Key parameters of the undersampling process, such as sampling period, power spectral density level, and type, are explored. Low-frequency noise and particularly 1/f noise impact show a major impact in results. The new calculation algorithm is implemented in MATLAB software. Two CMOS image sensor technologies are used to perform a comparison between noise estimation and measurements. This comparison shows a very good correlation between estimations and measurements.