On the optimal measurement of conversion gain in the presence of dark noise.

Working from a model of Gaussian pixel noise, we present and unify over 25 years of developments in the statistical analysis of the photon transfer conversion gain measurement. We then study a two-sample estimator of the conversion gain that accounts for the general case of non-negligible dark noise. The moments of this estimator are ill-defined (their integral representations diverge), and so we propose a method for assigning pseudomoments, which are shown to agree with actual sample moments under mild conditions. A definition of optimal sample size pairs for this two-sample estimator is proposed and used to find approximate optimal sample size pairs that allow experimenters to achieve a predetermined measurement uncertainty with as little data as possible. The conditions under which these approximations hold are also discussed. Design and control of experiment procedures are developed and used to optimally estimate a per-pixel conversion gain map of a real image sensor. Experimental results show excellent agreement with theoretical predictions and are backed up with Monte Carlo simulation. The per-pixel conversion gain estimates are then applied in a demonstration of per-pixel read noise estimation of the same image sensor. The results of this work open the door to a comprehensive pixel-level adaptation of the photon transfer method.

3D noise photon transfer curve.

A photon transfer curve (PTC) is used to determine fundamental detector noise parameters such as read noise, conversion gain, and fixed pattern noise. Here, the method for determining a PTC is expanded to include 3D noise parameters. 3D noise PTC provides more insight into detector noise and is treated as the next logical step to classical PTC. However, it induces several new challenges in analyzing the results, specifically the fitting of seven, or more, variance curves compared to the one (total variance) or two (temporal and fixed pattern variance) prior. Therefore, a general measurement model is created, which provides a new method to separate out all the classical terms, such as DSNU and PRNU, but can also handle high gain cameras with a noise factor. This method is then verified using Monte Carlo simulations and applied to a commercial machine vision camera. In addition, the effects of lens vignetting and non-uniformity correction (NUC) are explored, along with a comparison of the single pixel PTC.

Finite sampling corrected 3D noise with confidence intervals.

When evaluated with a spatially uniform irradiance, an imaging sensor exhibits both spatial and temporal variations, which can be described as a three-dimensional (3D) random process considered as noise. In the 1990s, NVESD engineers developed an approximation to the 3D power spectral density for noise in imaging systems known as 3D noise. The goal was to decompose the 3D noise process into spatial and temporal components identify potential sources of origin. To characterize a sensor in terms of its 3D noise values, a finite number of samples in each of the three dimensions (two spatial, one temporal) were performed. In this correspondence, we developed the full sampling corrected 3D noise measurement and the corresponding confidence bounds. The accuracy of these methods was demonstrated through Monte Carlo simulations. Both the sampling correction as well as the confidence intervals can be applied a posteriori to the classic 3D noise calculation. The Matlab functions associated with this work can be found on the Mathworks file exchange ["Finite sampling corrected 3D noise with confidence intervals," https://www.mathworks.com/matlabcentral/fileexchange/49657-finite-sampling-corrected-3d-noise-with-confidence-intervals.].