A single chip Photoplethysmography(PPG) sensor was developed for continuous measurements of heart rate from a mobile device. In order to utilize it in various mobile applications, it was necessary to achieve low power and small size of PPG sensors. For low power and small chip size of the PPG sensor, a photodiode(PD) for sensing signals and an analog front end(AFE) for signal amplification and ADC should be implemented in a single chip. The single chip PPG sensor which is implemented on a standard CMOS process with low operating voltage could be more suitable for mobile devices. In order to operate at a low voltage, reduction of Si thickness is required, and for this, high quantum efficiency(QE) of 43% at 940nm were obtained at 3um thickness by back side trench(BST) pattern and ARL optimization. In addition, to improve the performance of the PPG sensor, the leakage current of <0.1nA and capacitance of <200pF were measured by 20um pixel array. As a result, the low-power, small size single-chip PPG sensor showed similar performance to conventional high-voltage and large PPG sensor.
Can a mobile camera see better through display? Under Display Camera (UDC) is the most awaited feature in mobile market in 2020 enabling more preferable user experience, however, there are technological obstacles to obtain acceptable UDC image quality. Mobile OLED panels are struggling to reach beyond 20% of light transmittance, leading to challenging capture conditions. To improve light sensitivity, some solutions use binned output losing spatial resolution. Optical diffraction of light in a panel induces contrast degradation and various visual artifacts including image ghosts, yellowish tint etc. Standard approach to address image quality issues is to improve blocks in the imaging pipeline including Image Signal Processor (ISP) and deblur block. In this work, we propose a novel approach to improve UDC image quality - we replace all blocks in UDC pipeline with all-in-one network – UDC d^Net. Proposed solution can deblur and reconstruct full resolution image directly from non-Bayer raw image, e.g. Quad Bayer, without requiring remosaic algorithm that rearranges non-Bayer to Bayer. Proposed network has a very large receptive field and can easily deal with large-scale visual artifacts including color moiré and ghosts. Experiments show significant improvement in image quality vs conventional pipeline – over 4dB in PSNR on popular benchmark - Kodak dataset.
Under Display Camera(UDC) technology is being developed to eliminate camera holes and place cameras behind display panels according to full display trend in mobile phone. However, these camera systems cause attenuation and diffraction as light passes through the panel, which is inevitable to deteriorate the camera image. In particular, the deterioration of image quality due to diffraction and flares is serious, in this regard, this paper discusses techniques for restoring it. The diffraction compensation algorithm in this paper is aimed at real-time processing through HW implementation in the sensor for preview and video mode, and we've been able to use effective techniques to reduce computation by about 40 percent.
This paper documents the design, construction, and experimental evaluation of an ultra-low-cost large-format digital camera. Used lenses that cover formats up to 4x5 can be surprisingly inexpensive, but large-format image sensors are not. By combining 3D printing with cheap components developed for use in IoT (Internet of Things) devices, especially the sub-$10 ESP32-CAM, a digital scanning large-format camera capable of over 2GP resolution can be constructed at very low cost. Despite the large image area, Lafodis160 is literally a wireless IoT device, fully remote controllable via Bluetooth and WiFi. This camera was originally intended to serve as a testbed for novel ways to improve capture quality for scenes that are not completely static during the scan interval, and a brief overview is given of methods employing unusual scan orderings that will be evaluated using it.
In a mobile smartphone camera, image quality is more degraded towards the edges of an image sensor due to high CRA (Chief Ray Angle). It is critical to estimate the cause of this effect since image quality is degraded at image periphery from attenuating illuminance and broadening PSF (point spread function). In order to predict image quality from the center to the edge of the camera output, we propose a method to estimate lens PSFs at any particular image field. The method adopts Zernike polynomials to consider lens aberrations while having an arbitrary spatial sampling. Also, it employs estimating a pupil shape in accordance with an optical field. The proposed method has two steps: 1) estimation of a pupil shape and Zernike polynomial coefficients, and 2) generation of a PSF with estimated parameters. The method was experimented with a typical mobile lens to evaluate the performance of the PSF estimation at 0.0F and 0.8F. In addition, Siemens star images were generated with the estimated PSFs to compare resolutions at the center and the edge of an image. The results show that the image of the edge is worse than that of the center in terms of MTF (Modulation Transfer Function), showing the importance of assessing image quality at the edge for pre-evaluation of a mobile camera.
A low-voltage pixel with 0.7 µm pitch was designed for a low-power CMOS image sensor. By reducing a pixel power supply voltage (Vpix), power consumption for pixel was reduced, but full-well capacity (FWC) was also decreased. However, by lowering the conversion gain (CG) and applying a negative voltage to the ground (NGND) of the pixel, FWC of 6000 e- was achieved without any degradation of both charge transfer lags and backflow noise. In addition, the signal linearity in the reduced analog-to-digital (ADC) range was improved by optimizing the source follower (SF). For dark performances, white spots and dark current worsened by NGND were significantly improved by forcing more negative voltage to the transfer gate (TG) when it was turned off.
In this paper we propose a low power consumption high speed analog correlated multiple sampling (CMS) technique with high density switched capacitors for low noise CMOS image sensors. A CIS with 256 analog memories per pixel using high density trench capacitors was employed in order to verify the noise reduction effect dependent on operation timings. The noise characteristics were measured at sampling numbers of M=1~64 with various CMS sampling period of 10ns to 1µs and time interval between reset and signal samplings, thanks to the high flexibility owing to the proposed analog CMS technique. The measurement results agree well with the theoretical calculation results, showing that conducting CMS with highly correlated signals is effective in noise reduction.
Low optical-crosstalk color pixel scheme with wave-guiding structures is demonstrated in a high resolution CMOS image sensor with a 0.8um pixel pitch. The high and low refractive index configuration provides a good confinement of light waves in different color channels in a quad Bayer color filter array. The measurement result of this back-side illuminated (BSI) device exhibits a significant lower color crosstalk with enhanced SNR performance, while the better angular response and higher angular selectivity of phase detection pixels also show the suitability to a new generation of small pixels for CMOS image sensors.
LCDs (Liquid Crystal Displays) have become the ubiquitous low-cost display technology, with full color displays offering good resolution costing less than $10. Although LCD modules generally include either a backlight or a reflective backing, the LC panel itself merely modulates light by altering polarization. Thus, it is possible to use a transmissive LC panel as a programmable optical filter, or LCLV (Liquid Crystal Light Valve). This paper explores a variety of potential uses of commodity LC panels, including color panels, to implement programmable apertures and filters for camera lenses.