OmniVision Technologies

OmniVision's front-side illumination (FSI) technology is used to manufacture compact cameras in mobile handsets, notebook computers and other applications that require low-light performance without the need for flash.

OmniPixel3-GS expands on its predecessor, and is used for eye-tracking for facial authentication,[14] and other computer vision applications.

Backside illuminated image (BSI) technology differs from FSI architectures in how light is delivered to the photosensitive area of the sensor. In FSI architectures, the light must first pass through transistors, dielectric layers, and metal circuitry. In contrast, OmniBSI technology turns the image sensor upside down and applies color filters and micro lenses to the backside of the pixels, resulting in light collection through the backside of the sensor.

The second-generation BSI technology, developed in cooperation with Taiwan Semiconductor Manufacturing Company Limited (TSMC), is built using custom 65 nm design rules and 300mm copper processes. These technology changes were made to improve low-light sensitivity, dark current, and full-well capacity and provide a sharper image.

In this camera module, sensor and lens manufacturing processes are combined using semiconductor stacking methodology. Wafer-level optical elements are fabricated in a single step by combining CMOS image sensors, chip scale packaging processes, (CSP) and wafer-level optics (WLO). These fully integrated chip products have camera functionality and are intended to produce thin, compact devices.

RGB-iR technology uses a color filter process to improve color fidelity. By committing 25% of its pixel array pattern to infrared (IR) and 75% to RGB, it can simultaneously capture both RGB and IR images. This makes it possible to capture both day and night images with the same sensor. It is used for battery powered home security cameras as well as biometric authentication, such as gesture and facial recognition.[15]

OmniVision developed its PureCel and PureCel Plus image sensor technology to provide added camera functionality to smartphones and action cameras. The technical goal was to provide smaller camera modules that enable larger optical formats and offer improved image quality, especially in low-light conditions.[16]

Both of these technologies are offered in a stacked die format (PureCel-S and PureCelPlus-S). This stacked die methodology separates the imaging array from the image sensor processing pipeline into a stacked die structure, allowing for additional functionality to be implemented on the sensor while providing for much smaller die sizes compared to non-stacked sensors. PureCelPlus-S uses partial deep trench isolation (B-DTI) structures comprising an interfacial oxide, first deposited HfO, TaO, oxide, Ti-based liner, and a tungsten core. This is OmniVision's first DTI structure, and the first metal filled B-DTI trench since 2013.[17]

PureCel Plus uses buried color filter array (BCFA) to collect light with various incident light angles for tolerance improvements. Deep trench isolation reduces crosstalk by creating isolation walls between pixels inside silicon. In PureCel Plus Gen 2, OmniVision set out to improve deep trench isolation for better pixel isolation and low-light performance. Its target application is smartphone video cameras.[18]

Developed to address the low-light and night-vision performance requirements of advanced machine vision, surveillance, and automotive camera applications, OmniVision's Nyxel NIR imaging technology combines thick-silicon pixel architectures and careful management of the wafer surface texture to improve quantum efficiency (QE). In addition, extended deep trench isolation helps retain modulation transfer function without affecting the sensor's dark current, further improving night vision capabilities.[19] Performance improvements include image quality, extended image-detection range and a reduced light-source requirement, leading to overall lower system power consumption.[20]

This second generation near-infrared technology improves upon the first generation by increasing the silicon thickness to improve imaging sensitivity. Deep trench isolation was extended to address issues with crosstalk without impacting modulation transfer function. Wafer surface has been refined to improve the extended photon path and increase photon-electron conversion. The sensor achieves 25% improvement in the invisible 940-nm NIR light spectrum and a 17% increase in the barely visible 850-nm NIR wavelength over the first-generation technology.[21]

High-dynamic-range (HDR) imaging relies on algorithms to combined several image captures into one to create a higher quality image than native capture alone. LED lighting can create a flicker effect with HDR. This is a problem for machine vision systems, such as those used in autonomous vehicles. That is because LEDs are ubiquitous in automotive environments, from headlights to traffic lights, road signs and beyond. While the human eye can adapt to LED flickering, machine vision cannot. To mitigate this effect, OmniVision uses split-pixel technology. One large photodiode captures a scene using short exposure time. A small photodiode using long exposure simultaneously captures the LED signal. The two images are then joined in a final picture. The result is a flicker-free image.[22]

OmniVision CMOS image sensors range in resolution from 64 megapixels to below one megapixel.[23] In 2009, it received orders from Apple for both 3.2 megapixel and 5 megapixel CIS.[24]

OmniVision also manufactures application integrated circuits (ASICs) as companion products for its image sensors used in automotive, medical, augmented reality and virtual reality (AR/VR), and IoT applications.[25]

OmniVision's CameraCubeChip is a fully packaged, wafer-level camera module measuring 0.65 mm × 0.65 mm. It is being integrated into disposable endoscopes and catheters with diameters as small as 1.0mm. These medical devices are used for a range of medical procedures, from diagnostic to minimally invasive surgery. The used OV6948 sensor has a size of 0.575 mm × 0.575 mm and a resolution of 200 × 200 Pixel.[26]

OmniVision manufacturers liquid crystal on silicon (LCOS) projection technology for display applications.[27]

In 2018, AR start-up, Magic Leap, adopted both OmniVision LCOS technology and sensor bridge ASIC for its Magic Leap One mixed reality products.[28]