Chen Cao

Insufficient signal-to-noise ratio (SNR) limits the quality of cardiac pulses measured by multitap CMOS image sensors (CISs). We present an in-pixel temporal redundant samplings (TRS) method for enhancing the SNR through sampling the charges originated by one individual optical scene into the multiple storage taps in turn with the identical time window, enabling magnification of the image signal in shot noise domain. Modeling indicates that the linear SNR increases with a factor square root of TRS sampling number, which is also verified by characterization results utilizing a four-tap CIS. Application results show that the high-frequency (HF) noise components observed in the measured cardiac pulses can be attenuated by 67.9%, 48%, and 53.6% with heart rate detection accuracies up to 99.7%, 99.3%, and 98.7% by applying TRS modes under given conditions of stable ambient, background light (BGL) variation, and motion disturbance, respectively. Moreover, the proposed method is as well foreseeable to be extended to other multitap-involved imaging applications.

A 7.1- μm-pitch lock-in pixel (LIP) CMOS image sensor (CIS) that is capable of simultaneously phasing with two different near-infrared-band (NIR-band) light pulses is developed for remote physiological measurement, particularly a heart rate variability (HRV) signal monitor. The proposed dual NIR-band lock-in operation is realized by a four-tap-wise charge modulation with adjacent two-tap LIPs by means of a column-interdigital pixel driving scheme combining with in-LIP lateral electric field modulators (LEFMs). For attaining the weak HRV signal, the implemented CIS achieves a subelectron temporal noise of 0.67 e ^- _rms at 64 times correlated multiple sampling (CMS) and a high charge modulation contrast (MC) up to 96% at maximum. These remarkable properties primarily result from the in-LIP dark current and LEFM controlling optimizations proposed in device domain. The experimental results show that the developed CIS exhibits not only a 98.5% accuracy in HRV signal capture with -39.1-dB signal-to-background ratio (SBR) against a fluctuant brightness ambient light but also a moderate artificial motion robustness benefited from the proposed dual NIR-band lock-in technique.

This paper presents a time-resolved (TR) CMOS image sensor (CIS) using two-tap near-infrared (NIR) lock-in pixels with background light (BGL) cancelling capability for remote heart rate (HR) detection. Aimed to the application, the sensor features a 1024 (V) x 1280 (H) spatial resolution with 412 (V)x 1280 (H) region-of-interest-readout (ROI-readout) at 30 fps, a high sensitivity of 134.8 ke⁻/lux · s, and an ultra-low temporal random noise of 1.13 erms⁻ at analog gain of 64. A maximum charge modulation ratio of 90.1% is obtained for realizing lock-in operation by means of in-pixel lateral electric field charge modulator (LEFM). The remote HR detection is achieved by capturing the heart-beat-induced temporal variation of the oxy-hemoglobin (HbO₂) concentration by the developed TR CIS with NIR lock-in technique, and larger than 98% detection accuracies are attained under both non-visible-light condition and fluctuant brightness ambient-light environments. This remarkable feature demonstrates the superiority of the proposed method for driver monitoring applications in particular.

A time-resolved CMOS image sensor (CIS) using two-tap near infrared (NIR) lock-in pixels with background light (BGL) cancelling capability for non-contact heart rate (HR) detection is presented. Aimed to the application, the sensor performances of 1.1e ^- _rms temporal random noise and 90.1% maximum charge modulation ratio are achieved by using low noise readout circuit and in-pixel lateral electric field charge modulator (LEFM) techniques. The non-contact HR detection with accuracies of more than 98% is implemented by the developed sensor under both non-visible-light and fluctuant ambient-light brightness conditions. This remarkable performance indicates the superiority of the proposed technique for automotive security applications in particular.

The charge transfer potential barrier (CTPB) formed beneath the transfer gate causes a noticeable image lag issue in pinned photodiode (PPD) CMOS image sensors (CIS), and is difficult to measure straightforwardly since it is embedded inside the device. From an understanding of the CTPB formation mechanism, we report on an alternative method to feasibly measure the CTPB height by performing a linear extrapolation coupled with a horizontal left-shift on the sensor photoresponse curve under the steady-state illumination. The theoretical study was performed in detail on the principle of the proposed method. Application of the measurements on a prototype PPD-CIS chip with an array of 160 × 160 pixels is demonstrated. Such a method intends to shine new light on the guidance for the lag-free and high-speed sensors optimization based on PPD devices.

An improved analytical model for quantifying the full well capacity in pinned photodiode (PPD) CMOS image sensors is proposed. The model captures the characteristics of the realistic technology-induced vertical doping nonuniformity in photon sensing N-type area of the PPD structure and the voltage dependency of the PPD capacitance, respectively, both of which were neglected in the existing works. Excellent agreement between measured and predicted data shows that the proposed model fits a wider range of technology conditions for a wider spectra responding compared to the up-to-date reported model.