Top 10 image sensors

Delivering better image quality is a top challenge in all applications for image sensor manufacturers. Improvements in image sensors help design engineers enable advanced features for their automotive, machine vision, mobile device and security and surveillance designs. These new sensors deliver higher resolution, improved high dynamic range, increased pixel density and higher sensitivity. They also offer better low-light capabilities.

According to Global Market Insights Inc., the global image sensor market is expected to increase from $20 billion in 2022 to over $50 billion by 2032, with a compound annual growth rate of 5% from 2023 to 2032. CMOS image sensors still hold the lion’s share of the market, at greater than 85% in 2022.

One of the biggest growth drivers is the automotive sector. Automakers continue to add advanced features, such as lane-change detection, motion tracking and advanced driver-assistance systems (ADAS). They are also using high-resolution image sensors for features like autonomous driving (AD), as well as rear-view, front-view and surround-view monitoring, according to the market research firm. This includes the increased use of near-infrared (NIR) image sensors for improved vision in low-light environments.

Other growth drivers for image sensors, according to Global Market Insights, are developments in automation and robotics that call for high-resolution color sensing and enhanced vision pixels, rising demand for surveillance systems and increased smartphone production in the Asia-Pacific region.

Smartphones and other mobile devices are big drivers in the demand for higher-quality images, increasing the need for high-resolution cameras and thus more advanced image sensors that also consume less power in more compact packages.

Here are a few examples of image sensors, introduced in the past year, in each of these key market segments, offering enhanced performance for next-generation designs.

Automotive

Onsemi has unveiled its Hyperlux automotive image sensor family with ultra-high dynamic range (HDR) for ADAS and AD solutions. The image sensors, with a 2.1-µm pixel size, claim industry-leading 150-dB HDR and LED flicker mitigation (LFM) across the full automotive temperature range.

The Hyperlux family spans products with resolutions from 3 megapixels (MP) to 8 MP and higher, for both sensing and viewing camera applications. Thanks to the HDR of 150 dB, the Hyperlux sensors can capture high-quality images under the most extreme lighting conditions without sacrificing low-light sensitivity, while the LFM capability ensures that pulsed light sources do not appear to flicker, avoiding flicker-induced machine-vision issues, onsemi said.

The Hyperlux product family also claims the industry’s lowest power consumption and smallest footprint. The image sensors use up to 30% less power with up to a 28% smaller footprint than competing devices for energy-efficient designs. The sensors also save space, which is key to vehicle design as more electronic content is added to cars, the company said.

Hyperlux meets the requirements of Automotive Safety Integrity Level D (ASIL-D) systems, while integrated security provides protection against cybersecurity threats.

These sensors offer a dual output stream at different resolutions, allowing sensing and other functions to be performed simultaneously. The flexible architecture enables the use of a single camera for multiple functions, reducing design time, risk and cost, onsemi said.

The first two Hyperlux sensors released are the AR0823AT and AR0341AT. They are now sampling to early-access customers.

Also targeting automotive safety applications is STMicroelectronics’ hybrid image sensor. The new sensor is designed to extend driver-monitoring systems (DMS) beyond assessing driver alertness to full interior monitoring of the driver and all passengers. The new hybrid rolling and global-shutter image sensor will enable new applications, such as passenger safety-belt checks, vital-sign monitoring, child-left detection, gesture recognition and high-quality video/picture recording.

The new sensor, the VD/VB1940, combines the sensitivity and high resolution of infrared sensing with HDR color imaging in a single component. The hybrid sensor can capture frames alternatively in rolling-shutter and global-shutter modes.

The 5.1-MP image sensor in a 1/2.5-inch optical format captures HDR color images for occupant-monitoring systems and NIR images typically captured by standard DMS sensors. NIR imaging can analyze driver head and eye movements in all lighting conditions.

The sensor captures NIR images in global-shutter mode, allowing synchronization with the infrared LED emitter to capture fast-moving scenes without motion blur, ST said. In rolling-shutter mode, which reads pixel data row by row, the VD/VB1940 is said to deliver optimized color-imaging performance. It delivers up to 60 frames per second (fps) at full resolution.

The automotive image sensor is built on ST’s second-generation 3D-stacked backside-illuminated (BSI) wafer technology. This maximizes the optical area and on-chip processing in relation to die size and allows the sensor to perform sophisticated algorithms locally for optimal performance in color and NIR imaging, ST said. This results in power savings and eases demand for an external co-processor.

Algorithms performed on-chip include Bayer conversion and HDR merging. With the on-chip HDR-merge feature in rolling-shutter mode, the VB/VD1940 is reported to produce a 100-dB full-resolution color image.

ST said the “on-chip Bayerization processing enables the user to reshuffle the color pixels of the RGB NIR 4 × 4 pattern into RGGB format,” which is compatible with a variety of SoCs. The embedded processor manages cybersecurity features to ensure privacy in connected-vehicle applications, including mutual authentication and pairing of camera and electronic control unit, as well as video-stream authentication.

The hybrid sensor is available in bare wafers (VDB1940) and packaged in BGAs (VB1940). Samples are available, with mass production planned for 2024 vehicles. The sensor is qualified to AEC-Q100 and is ISO 26262–compliant for use in functional-safety systems up to ASIL-B.

STMicroelectronics also offers an advanced global-shutter image sensor for DMS designs. The second-generation design leverages ST’s 3D-chip technology and in-house investment in manufacturing advanced 3D-stacked BSI 3D image sensors, which are more sensitive, smaller and more reliable than conventional frontside-illuminated sensors typically used in first-generation DMSes, ST said.

The VB56G4A achieves high quantum efficiency (QE), reaching 24% at 940-nm NIR wavelength, with linear dynamic range up to 60 dB, enabling a simple, low-power, non-visible LED emitter to provide adequate illumination for the sensor, ST said.

The sensor’s high QE, together with a pixel size of 2.6 µm, helps optimize total power consumption and camera size, the company said.

Other features include integrated automatic exposure control and flexible operating modes that help optimize system features and performance, such as programmable sequences of four-frame contexts, illumination control outputs synchronized with sensor integration periods, an input for an external frame-start signal, automatic dark calibration, dynamic defective-pixel correction, image cropping and a mirror/flip-image readout.

External connections include eight programmable general-purpose I/O pins and a dual-lane MIPI CSI-2 transmitter interface operating up to 1.5 Gbits/s per lane. The sensor can operate at up to 88 fps at full resolution, and typical power consumption is 145 mW at 60 fps.

Mass production is scheduled for the beginning of 2023, with adoption in model-year 2024 vehicles.

Machine vision

Omnivision recently introduced a high-sensitivity global-shutter image sensor, addressing the need for small and low-power–consuming cameras in AR/VR/MR and metaverse applications as well as drones, machine vision and barcode scanners.

The OG0VE image sensor features a sensitive 3.0-micron pixel based on the company’s OmniPixel3-GS technology and is 26% smaller and more than 50% more power-efficient than its predecessor.

Designed for any application requiring simultaneous localization and mapping, gesture detection, head and eye tracking, and depth and motion detection, the global-shutter image sensor provides a 640 × 480 resolution and pixel size of 3.6 × 2.7 mm in a 1/7.5-inch optical format. It features low power consumption: <34 mW when running at 60 fps with VGA.

The sensor features automatic black-level calibration and programmable controls for frame rate, mirror and flip, cropping and windowing. Other specifications include support for horizontal and vertical 2:1 and 4:1 monochrome subsampling and 2 × 2 monochrome binning. It also embeds 256 bits of one-time–programmable memory for part identification and programmable I/O drive capability. The operating temperature range is –30°C to 70°C.

The sensor supports one-lane MIPI D-PHY at up to 800 Mbits/s and is available in Omnivision’s CSP package or as bare die. The OG0VE global-shutter image sensor is available for sampling now and will be in mass production in mid-2023.

Gpixel Inc. recently announced a high-speed version of its GMAX4002, a 1/1.7-inch–format global-shutter CMOS image sensor, based on a 4-µm charge-domain global-shutter pixel with enhanced NIR sensitivity and a 2,048 × 1,200 resolution.

The high-speed version of the 2.4-MP GMAX4002 sensor supports a 50% increase in the data and frame rate achieved by each of the 8 sub-LVDS channels, as well as an increased bit rate from 800 Mbits/s to 1,200 Mbits/s per channel in 10-bit mode. The total 9.6-Gbits/s data rate makes it suited for emerging industry-standard camera interfaces like 10 GbE, Gpixel said.

This results in the full-resolution frame rate of 344 fps, while maintaining over 64 dB of dynamic range and less than three electrons noise, the company said, and an additional increase in the frame rate can be achieved by selecting a vertical region of interest for readout. The higher frame rate makes the sensor suited for machine vision, including in-situ optical inspection applications like PCB and wirebond assembly inspection, 3D inspection applications, motion capture and analysis, as well as sports, broadcast and entertainment applications.

The high-speed GMAX4002 maintains a peak QE of 72% at 550 nm and over 37% at 850 nm, thanks to Gpixel’s proprietary RedFox NIR technology. In addition, sensitivity can be increased and frame rate quadrupled to over 680 fps using the sensor’s on-chip, charge-domain 2 × 2 binning mode, Gpixel said.

The high-speed version of the GMAX4002 is pin- and footprint-compatible with the standard-speed version. Both are packaged in a 74-pin CLCC package with outer dimensions of 16.6 × 12.9 mm and are available in Bayer color and monochrome versions. Devices and evaluation kits are available.

Mobile devices

Ams Osram expanded its Mira family of pipelined, high-sensitivity, global-shutter CMOS image sensors with the launch of the compact Mira050. Designed to save space and power in mobile devices and wearables, including smart glasses and VR headsets, the new 0.5-MP image sensor, in a 2.3 × 2.8-mm CSP, is highly sensitive to both visible and NIR light.

The NIR image sensor incorporates numerous ultra-low–power features. Power consumption is 47 mW at full resolution and a frame rate of 120 fps, and standby power consumption is very low at 60 µW. The combination of the low power consumption and high sensitivity requires less illumination, which helps extend battery runtime, the company said.

The image sensor also incorporates on-chip power management that adjusts the power supplied to various functional blocks in response to settings for the frame rate and exposure time, which further helps to extend battery runtime. Target applications include eye tracking, gesture tracking and contextual awareness in AR/VR/MR headsets, object detection in robots and 3D depth sensing for face recognition in smart door locks.

The high sensitivity is also matched by high QE from 36% at 940 nm and 56% at 850 nm in the NIR spectrum and up to 93% at 550 nm in the visible light range, based on the company’s internal testing. The benefit is extra power savings because the sensor can operate with a low-power illuminator and in naturally low-light conditions, the company said.

The Mira image sensor family uses BSI technology to implement a stacked-chip design. The BSI technology enables high sensitivity and QE with a pixel size of 2.79 μm. Effective resolution of the CSP is 576 × 768 pixels and the maximum bit depth is 12 bits, with a bare-die version at 600 × 800 pixels.

Features include programmable registers that enable the control of window coordinates, timing parameters and exposure time, and mirror, flipping and cropping functions, as well as a MIPI CSI-2 interface for easy interfacing with a processor or FPGA. The on-chip registers can be accessed via a camera control interface for easy configuration of the sensor, the company said.

In January, Samsung Electronics Co. Ltd. released its latest 200-MP image sensor, the ISOCELL HP2, with improved pixel technology and full-well capacity for high-quality mobile images for premium smartphones. The ISOCELL HP2 boosts pixel capacity with 200 million 0.6-μm pixels in a 1/1.3-inch optical format, which is a size that is widely used in 108-MP main smartphone cameras, Samsung said, delivering higher resolutions without an increase in camera size.

A key feature of the ISOCELL HP2 is Samsung’s advanced pixel-binning technology, Tetra²pixel, which simulates different pixel sizes to accommodate different lighting levels. Here’s how it works: In low-lit environments, the sensor transitions to either a 1.2-μm 50-MP or 2.4-μm 12.5-MP image sensor by binding four to 16 neighboring pixels.

For fuller 8K video, approximately at 33 MP, the HP2 switches to 1.2-μm 50-MP mode to minimize cropping and capture more of the scene, explained Samsung. “Filming 8K at 30 fps, a wide field of view along with bigger pixel size can produce sharp cinematic videos.”

The ISOCELL HP2 also incorporates Samsung’s new Dual Vertical Transfer Gate (D-VTG) technology that significantly reduces washed out images from brightly lit environments. The D-VTG adds a second voltage transfer gate in the pixel (in a photo diode within each pixel) to boost the pixel’s full-well capacity by more than 33%. This reduces overexposure and enhances color reproduction, especially in bright-light conditions, the company said.

In low-lit settings, the ISOCELL HP2 uses super quad phase detection (QPD) for faster and more accurate auto-focusing. It allows the sensor to use all of its 200 million pixels for focusing agents, which are grouped by four adjacent pixels to recognize both horizontal and vertical pattern changes. The pattern data together with the number of reference points makes the sensor capable of fast auto-focusing, even in a dimly lit environment, Samsung said.

In addition, Samsung introduced its dual-slope–gain feature for the first time in 50-MP mode for excellent HDR performance. The company also includes its Smart-ISO Pro HDR solution that allows the camera to take 12.5-MP images and 4K at 60-fps video in HDR. The Samsung ISOCELL HP2 is in mass production.

Samsung Electronics also introduced the 200-MP ISOCELL HP3 image sensor with a 0.56-μm pixel for smartphones. Claiming the industry’s smallest pixel, at launch in June 2022, the upgraded ISOCELL HP3 image sensor is expected to deliver ultra-high resolution in smartphone cameras.

The ISOCELL HP3, with a 12% smaller pixel size than the predecessor’s 0.64 μm, delivers 200 million pixels in a 1/1.4-inch optical format, translating into about a 20% reduction in camera-module area, Samsung said.

The ISOCELL HP3 features super QPD auto-focusing, which uses a single lens over four adjacent pixels to detect the phase differences in both horizontal and vertical directions. This enables a more accurate and quicker auto-focusing for smartphone camera users, the company said.

The sensor supports videos in 8K at 30 fps or 4K at 120 fps. Samsung said there is a minimal loss in the field of view when taking 8K videos, and when combined with the super QPD solution, it creates movie-like cinematic footage.

It also uses the Tetra²pixel technology for brighter and more vibrant images even in low-light conditions or at night. The technology combines four pixels into one to transform the 0.56-μm 200-MP sensor into a 1.12-μm 50-MP sensor, or a 12.5-MP sensor with 2.24-μm pixels by combining 16 pixels into one, the company said.

The ISOCELL HP3 also uses the Smart-ISO Pro feature to maximize the dynamic range of the mobile image sensor. The technology merges image information from the two conversion gains of low and high ISO mode to create HDR images. The technology comes with a triple ISO mode — low, mid and high — that further widens the sensor’s dynamic range. It also supports over 4 trillion colors (14-bit color depth), 64× more colors than the predecessor’s 68 billion (12-bit color depth).

In addition, by supporting staggered HDR together with Smart-ISO Pro, the ISOCELL HP3 can switch between the two solutions depending on the environment to produce high-quality HDR images, Samsung said.

Security and surveillance

Onsemi packs a number of innovative features into its new AR0822 image sensor for security and surveillance applications. These include 120-dB embedded HDR (eHDR), high-efficiency NIR response needed for applications in harsh lighting conditions and several power-saving features, including a low-power architecture and wake-on-motion, which allows the processor to enter a low-power standby mode until the sensor detects motion and triggers the processor to resume operating mode. This sensor can also be used in body cameras, doorbell cameras and robotics.

The AR0822 is an 8-MP stacked 1/1.8−inch (8.81-mm diagonal) BSI CMOS digital image sensor based on a 2.0-µm pixel that can achieve 4K video quality. It features an active-pixel array of 3,840 × 2,160 and is capable of 4K video at 60 fps, capturing images in either linear or eHDR modes (120 dB) with a rolling-shutter readout.

Onsemi explained that with eHDR functionality on the sensor, the AR0822 reduces system bandwidth and processor power and delivers high image quality, while compensating for motion and flickering-light sources. This is in comparison to many HDR techniques that use multi-exposure outputs that require more system bandwidth and more expensive components with higher resolutions, the company said.

The AR0822 also offers enhanced NIR sensitivity and sophisticated camera functions like binning/windowing. It is designed for rugged environments, with an operating temperature range from –30°C to 85°C junction temperature.

Sony Semiconductor Solutions Corp. also offers a CMOS image sensor targeting security cameras. When the IMX675, a 1/3-type image sensor with approximately 5.12 MP, was announced in 2022, the company called it the industry’s first CMOS image sensor for security cameras that simultaneously delivers both full-pixel output of the whole captured image and high-speed output of regions of interest.

Delivering high-quality image capture regardless of usage conditions or time of day, the IMX675 sensor uses dual-speed streaming technology to output all the pixels in a captured image at a maximum rate of 40 fps, while simultaneously outputting specific user-set regions of interest at high speed. This allows a single camera to provide comprehensive images of scenes and support high-speed recognition of specific objects at a high level of detail, while helping to reduce system costs, Sony said.

Usage examples cited include a whole image capture of an intersection while recognizing people and cars moving through it, or the capture of an overall image of a highway while also capturing the details needed to identify the license plate numbers of cars traveling on it.

The new sensor also uses the company’s unique STARVIS 2 technology, which delivers high sensitivity and a wide dynamic range of 78 dB, approximately 2.5× that of previous modules when used in a single exposure method, as well as approximately 30% less power consumption than previous models, thanks to its proprietary stacked structure. These features also make the image sensor suited for other applications, such as drive recorders and webcams.