Statistical grey level and noise evaluation of Foveon X3 and CFA image sensors

Abstract

This paper presents a practical, comprehensive and flexible statistical framework to analyze the radiometric response values and their uncertainty effects on image-based sensor systems. The statistical framework is based on experimental techniques and is tested on two single lens reflex (SLR) cameras powered by different image sensors, Foveon X3 and Color Filter Array (CFA). A colorchecker inside an indoor light booth is used as target. Design of Experiments (DoE) for Linear Models (LM) are derived to analyze and characterize the variability of the output grey level signal (digital number) and their uncertainty effects according to influential factors such as scene reflectance, wavelength range and time. The digital numbers are statistically modeled, and their noise and temporal trend are the uncertainty effects analyzed. Experiments are carried out under laboratory conditions to minimize the rest of uncertainty sources that might affect digital numbers. The flexibility of the statistical framework is confirmed to model and characterize the digital numbers, as well as the noise of a single image and the stability (trend and noise) of a temporal sequence of images.