Reduced Human Perception of FPN Noise of the Pyramidal Readout CMOS Image Sensor

摘要

A novel CMOS active pixel sensor imager, called pyramidal readout architecture, is presented in which the 1-D row sampling of a classical imager is replaced by a 2-D ring sampling and diagonal output buses. The first characteristic resulted in enhancement of the readout speed of the imager, while the second characteristic is shown here to make fixed pattern noise (FPN) less perceptible to the human visual system (HVS). This observation exploits the relatively poor sensitivity of the HVS to obliquely distributed patterns, known as the oblique effect. The effect was modeled using a spatial filter derived from measured characteristics of the HVS, and applied to the obliquely distributed FPN-dominated images of the pyramidal CMOS imager.