Reduction of random telegraph signal (RTS) noise in CMOS image sensors using histogram analysis

The paper presents column parallel signal processing techniques for reducing Random Telegraph Signal (RTS) noise of in-pixel source follower by using histogram analysis for the development of a very low-noise CMOS image sensor. In this method, a histogram with multiple samples for reset level is used to estimate the amplitude of the RTS noise. With the median of the histogram and the estimated amplitude, the RTS noise components are removed and the average is calculated with the histogram due to thermal noise only, to further reduce the noise level. Result of the application of the histogram-based noise suppres-sion to an implemented CMOS image sensor prototype for a large sampling numbers is demonstrated.