A Novel Hexagonal Photosite Grid

Abstract

Digital imaging has long been dominated by the use of square-grid photodiode arrays overlaid with a Bayer Color Filter Array (CFA). While this configuration has proven effective, it possesses inherent limitations in quantum efficiency, color fidelity, and effective spatial resolution due to light loss from trench/walls and the need for inefficient demosaicing algorithms. This paper introduces a novel sensor architecture featuring a hexagonal honeycomb grid of photodiodes with a corresponding red, green, and blue color filter pattern. We present a theoretical analysis demonstrating the advantages of this hexagonal geometry. Our findings indicate that the hexagonal array offers superior packing density, leading to a higher area to perimeter ratio and a theoretical increase in light sensitivity. Furthermore, the sampling lattice of a hexagonal grid provides a more isotropic response and improved angular resolution under demosaicing, which results in more accurate color interpolation and a significant reduction in aliasing artifacts such as color moiré. We project that this architecture can achieve a higher modulation transfer function (MTF) for a given photodiode count compared to traditional Bayer sensors, ultimately leading to images with superior detail and color rendition, particularly in photon-limited conditions, as well as a higher effective dynamic range due to an increased neighborhood availability of red and blue color inputs versus a bayer sensor.