Eye Tracker With In-Pixel Signal Processing for Fast Eye Tracking

This study proposed an eye tracker with in-pixel signal processing. The eye tracker achieved low-power consumption and latency by realizing a detection unit within a pixel, eliminating the need for image storage memory and interface circuit for image transmission. The detection unit utilized morphological operations—shrinkage and expansion—to detect a pupil or glint center (GC) and was implemented with all digital circuits, further improving the power efficiency. Additionally, an adaptive thresholding method was proposed to optimize latency for a given illumination condition. A threshold voltage for image binarization and integration time (IT) for a photodiode (PD) was adjusted to minimize the latency without compromising detection accuracy. The eye tracker comprising a <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$176\times 120$ </tex-math></inline-formula>-pixel array was fabricated using a 180 nm standard complementary metal-oxide semiconductor (CMOS) process and validated through appropriate experiments. It achieved a <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\surd 2$ </tex-math></inline-formula> pixel error with a latency of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$80~\mu $ </tex-math></inline-formula> s and dissipated 5 mW from a 3.3 V supply voltage.