Fixed pattern noise correction for wide dynamic range linear-logarithmic pixels

Pixels with the capability of linear response in low light regions and logarithmic response at high intensity have been proposed previously. A model for the complex response of these pixels has been devised that can be used to study the origins of fixed pattern noise and develop a fixed pattern noise correction procedure. The problem with correcting the fixed pattern noise in these pixels is the transition region between the linear and logarithmic operating regions. One way to avoid the problems from the transition region is to capture two outputs at different integration times within the same exposure. The resulting data can then be corrected for fixed pattern noise and converted to an equivalent linear or logarithmic response for subsequent image processing.