Monday, May 21, 2018

Hamamatsu Sensors in Automotive Applications

Hamamatsu publishes a nice article "Photonics for advanced car technologies" showing many applications for its light sensors:

Samsung Presentation

Samsung System LSI Investor Presentation dated by April 30, 2018 shows the company success in image sensor business:

  • 1/3 Global Smartphones use ISOCELL image sensors
  • 4.6 out of 10 Chinese smartphones use ISOCELL sensors
  • 28nm image sensor process

Sunday, May 20, 2018

Anafocus Keynote at EI 2018

Electronic Imaging Symposium publishes a keynote "Sub-Electron Low-Noise CMOS Image Sensors: Large Format, Fast, 0.5e-rms CIS with Oversampled 2‐Stage ADCs" by J. A. Segovia, F. Medeiro, A. Gonzales, A. Vellegas, and A. Rodriguez-Vazquez, Teledyne-Anafocus and Universidad de Sevilla.

Saturday, May 19, 2018

Omnivision Keynote at EI 2018

Electronic Imaging publishes Omnivision keynote presentation "Automotive Image Sensors" by Boyd Fowler and Johannes Solhusvik. The presentation covers many areas from HDR to LiDARs:

Friday, May 18, 2018

Imec and Holst Centre Unveil Organic Fingerprint Sensor

Imec: Holst Centre, a joint initiative of imec and TNO, have demonstrated a new class of flexible, large-area sensor technology for detecting finger- and palmprints. At less than 0.2 mm thick, the new sensors can be embedded into objects such as mobile phones and door handles to create “invisible” yet secure access control systems that can tell if the print is from a living person rather than a phantom or counterfeit.

Two demonstrators are shown: Measuring 6 x 8 cm, a 200-ppi demonstrator is large enough for 4-finger scanners that are currently used by border control authorities and delivers sufficient image quality for basic identification applications. Meanwhile, a slightly smaller 500-ppi demonstrator offer even higher image quality, compatible with FBI standards and enough for law enforcement agencies to visualize minutia and pores for more robust identification.

As with Holst Centre’s earlier flexible X-ray detectors, the fingerprint sensors combine an organic PD frontplane, an oxide TFT backplane, and a thin-film barrier for protection against the environment. All three technological elements have been or are being transferred to industry for scale-up and commercialization. The sensors read the finger- or palm print by detecting visible light (400 to 700 nm) reflected from the surface of the skin. However, they can also detect light that penetrates someway into the skin before being reflected. This allows them to sense a heartbeat from changes in the capillaries within the hand, and thus verify that the print comes from a live person.

By using different PD materials, the sensors’ band can be extended to other wavelengths such as NIR. This could enable new identification verification modes based on, for example, the pattern of veins in a hand, which is believed to be even more specific to an individual than a fingerprint.

The flexible fingerprint sensor demonstrator shows the versatility and maturity of the flexible electronics technologies that Holst Centre is developing. With the underlying technology already in use in the flat-panel industry, there is a fast route to manufacturing and we are looking for industrial partners to take that step,” says Hylke Akkerman, Program Manager at Holst Centre.

ST Focuses on 3D Imaging

ST Micro 2018 Capital Markets Day was used to present the company focus on all types of 3D imaging: SPAD-based FlightSense, structured light (probably Apple Face ID IR sensor), and also indirect ToF:

Thursday, May 17, 2018

Machine Learning to Reduce High-ISO Noise

MSN, DPReview: Machine learning is often presented as a tool to improve high-ISO. Intel and the University of Illinois Urbana-Champlain are to present their AI algorithm in June:

"The Intel and UIUC team claims the algorithm can now amplify low-light images the equivalent of up to 300 times the exposure, without the same noise and discoloration that programs like Photoshop might introduce or having to take two separate images."



Imaging Resource publishes an interview with Sony managers called "A.I. is the future of mirrorless cameras:"

"That kind of technology is improving very fast. For example, our current processor has some general computing unit, but maybe in the next generation, the general computing CPU should become more designed for AI."

In a half-year old video, Sony Teruo Hirayama and Sony’s strategic advisor also the director of the MIT Media Lab, Joichi Ito discuss how image sensors and AI will combine to enhance human capabilities and open up new possibilities in the future:

Omron MEMS Thermal Sensor

Omron publishes a video demoing its low-resolution thermopile sensor:

Wednesday, May 16, 2018

Framos Offers Sony SLVS-EC RX as FPGA IP

Framos SLVS-EC RX IP Core shortens design time with the latest generation of Sony image sensors. The SLVS-EC interface (Scalable Low-Voltage Signaling with Embedded Clock) has up to 8 lanes and 2.376 Gbps throughput per lane. When compared to Sony’s 2nd-generation CMOS image sensors with S-LVDS interfaces, the SLVS-EC interface doubles the overall output speed to 19 Gbps. SLVS-EC has more than three times higher bandwidth per lane.


Thanks to RP for the link!

Industry is Watching Market Response on Huawei Triple Camera

Digitimes: If Huawei is doing well with its P20 Pro smartphone with triple camera, other smartphone makers like Xiaomi, Oppo, Vivo, Lenovo, Asustek, HTC, and Sony are likely to come with similar solutions too, according to Digitimes sources. Huawei reportedly plans to extend its triple-lens cameras to its other high-end smartphone models.

Tuesday, May 15, 2018

iniVation Launches its Next-Generation Event-Based Dynamic Vision Sensor

iniVation, a spin-off of the University of Zurich and the ETH Zurich, launches its next-generation event-based DAVIS346 sensor with USB3 interface. It features over twice the resolution and a 30% smaller compared with previous models. The new sensor features:

  • Sub-millisecond response time
  • Ultra-high energy efficiency (>90% system-level improvement compared with conventional vision)
  • Wide DR >100 dB
  • Simultaneous APS frame and event capture
  • Small form factor (40 x 60 x 25 mm)
  • USB3 connector
  • Fully isolated GPIO ports for multi-camera sync

Leti on 3D Integration

December 2017 Leti presentation on 3D integration at IEDM Workshop talks about a number of opportunities for CIS:

Monday, May 14, 2018

Omnivision Announces Design Win in Automotive Eye and Gaze Tracking Camera

PRNewswire: OmniVision collaborates with Smart Eye to integrate its automotive 2MP image sensor with Smart Eye's sensing algorithm library in a turnkey driver and cabin monitoring solution that targets automotive OEMs.

"Automotive OEMs need image sensors that are certified to ASIL B for level-3 autonomous vehicles, beginning with 2021 models," said Marius Evensen, senior automotive technical marketing manager at OmniVision. "The OV2311's ASIL B certification includes advanced safety features."

Eye- and gaze-tracking cameras need to be small and unobtrusive so that drivers do not feel they are being monitored, but smaller cameras tend to run hot and create excess system noise. The OV2311 sensor in a chip-scale package consumes half the power of competing 1MP solutions, keeping the temperature under control without sacrificing performance. Its high pixel density enables a wide field of view for a larger headbox to adjust for various driver seat positions, along with precise images for tracking gaze. Additionally, the wider field of view opens the opportunity for sensing throughout the cabin interior, including passengers as well as the driver.

"OmniVision's OV2311 is now the benchmark for the driver monitoring industry. Our combined solution benefits automotive OEMs needing a complete hardware and software package that has been thoroughly tested and is ready to implement in a wide variety of vehicles," said Karolina Wikander, head of business management at Smart Eye.

LFoundry 2017 Report

LFoundry 2017 report contains interesting info about the foundry, its customers, and the relationships with its mother company SMIC:

"The year began with the clear, challenging objective of equipping the Avezzano plant with further, alternative production capacity compared to its traditional focus on image sensors, made necessary by the growing demand from China and Asian markets.

This demand, however, then saw a sudden turnaround, and, at the same time, a strong increase in the request for image sensors on the part of our largest, long-standing customer. Therefore, we spent the second part of 2017 involved in another important company reorganisation process, undertaken in order to guarantee this long-established customer sufficient supplies to meet their growing demand for image sensors.

Today, LFoundry is an increasingly strong candidate to become a SMIC centre of excellence for CMOS Image Sensors, and to take on the role of strategic partner of the world leaders in the application of image sensors in the automotive field.
"

More Info on SensL Acquisition by ON Semi

The Irish Times times publishes an article on ON Semi acquisition of SensL:

"Financial details of the transaction have not been disclosed but industry insiders told The Irish Times it was a “high double-digit deal”. The Bank of Ireland-backed Delta fund had a 45 per cent stake in the company prior to the acquisition with management and employees owning the bulk of the remaining shares.

SensL, whose clients include the European Space Agency, was spun out of the Tyndall National Institute in 2004. It was founded by Carl Jackson and Joe O’Keeffe, the latter of whom went on to enjoy success as a founder of Infiniled, which was acquired by Facebook-owned Oculus VR in 2016. SensL, which is now led by Bryan Campbell, employs 25 people in Cork.

“We built the company making optical sensors that can be used in a whole bunch of different products,” Mr Jackson told The Irish Times.

“We built the business on the back of our involvement in medical imaging, where we developed a key sensor that goes into PET scanners. We’re also active in radiation detection, and three years ago started working on lidar products for the automotive market. This will be one of our core focuses going forward,” he added.

While SensL’s autonomous car technology was one big attraction for On Semiconductor, Mr Jackson stressed that the buyer was interested in all of its product lines.
"

Update: The Sunday Times reports "Tech sources said ON Semiconductor was paying €60m-€70m cash for Sensl, which was founded in 2004. The takeover will deliver a significant return for Sensl’s shareholders, which include Irish venture capitalists Delta Partners and Enterprise Equity."

Friday, May 11, 2018

Albert Theuwissen Lectures on Video

CEI makes available for purchase 3 videos with Albert Theuwissen lectures:

Introduction to Correlated and Uncorrelated Noise in Imagers

Characterization of Noise in Dark

Characterization of Noise with Light

Oppo 3D Sensing Update

Mydrivers: A number of Chinese news sites cover Oppo May 10th meeting devoted to the company work on the 3D sensing for smartphones. Unfortunately, most of the information is in Chinese, but one can understand that Oppo is preparing structured light based 3D face recognition and in many other applications too. The 3D camera is said to be mass production ready and Oppo plans to launch first phones with it in the next 6 months:

Physically Unclonable Function based on CMOS Sensor FPN

University of Wisconsin–Madison publishes a video "CamPUF: Physically Unclonable Function based on CMOS Image Sensor Fixed Pattern Noise" by Yongwoo Lee, Kyuin Lee, and Younghyun Kim presented at Design Automation Conference (DAC) 2018:

EMVA1288 Characterization Video

Spectronet publishes a presentation of Bernd Jähne from Universität Heidelberg on image sensor characterization with some interesting findings on PRNU wavelength dependence, QE, angular dependencies, interesting MTF shapes, etc:

Thursday, May 10, 2018

Yonhap on Sony vs Samsung Competition

Korean news agency Yonhap gives its view on Sony and Samsung market shares, also quoted by Business Standard and MSN:

"Sony and Samsung currently lead the global market for CMOS image sensors, with the Japanese rival holding a slight edge.

In 2016, Sony and Samsung took up 25.6 percent and 22.6 percent of the market, respectively, industry data showed.

The two companies held 28.3 percent and 25.4 percent, respectively, last year. SK hynix Inc., Samsung's smaller South Korean rival, accounted for roughly 8 percent to 9 percent over the cited period.

Industry watchers, however, claim Sony takes up 50 percent of the market in terms of sales, hovering far above Samsung's 20 percent.
"

Himax Reports Q1 Results

GlobeNewswire: Himax updates on its imaging business with the new focus on 3D imaging and new product - under-display optical fingerprint sensors:

"Himax SLiMTM 3D sensing solution for premium Android smartphone is mass production ready in Q1 2018 as planned, awaits smartphone makers and their ecosystem partners to integrate 3D sensing technologies into the new end products. Each smartphone makers have different design and requirement for facial recognition and online payment, such developments are taking longer than the Company anticipated. Company now targets the end of 2018 for shipment to customers

..we are pleased to report an emerging optical fingerprint opportunity for CMOS image sensor business. Combining the leading fingerprint solution design of our partner and a low-power CMOS image sensor with superior sensitivity which we fully customized for this purpose, a number of design-in projects are already ongoing and we expect more to come.

Along with the new smartphone designs featuring edge-to-edge displays, structured light 3D sensing and ASC 3D sensing, both for facial recognition, as well as under-display optical sensor for fingerprint authentication are three new approaches for phone unlock and online payment to replace the traditional home button with capacitive fingerprint technology. Himax is uniquely positioned in that we offer critical technologies in all of the three solutions and are already a key player by forming different collaboration partnerships for each of the three alternatives.
"

From the company's factsheet:

Subaru Adds Face Recognition to its SUV

Nikkei: Subaru adds its first "driver monitoring system (DMS)" to the US model of the "Forester" mid-size SUV. While other companies have similar systems for years, Subaru adds facial recognition for pre-registered drivers and automatically adjusts the position of the seat, angles of the side mirrors, the settings of the air conditioner, etc. for the first time in the world. The driver face can be identified even in the dark by using a NIR LED and camera.

ON Semi Expands Assembly & Test Facility in Rochester

BusinessWire: ON Semi expands its manufacturing facility in Rochester, NY. The site develops and manufactures image sensors for commercial, industrial and professional imaging applications, including machine vision, surveillance, traffic monitoring, medical and scientific imaging, and photography. Located on a 4.2-acre site with over 260,000 square feet of building space, the expanded facility supports wafer fab, wafer probe, assembly, and test and packaging operations for specialized high-performance CCD and CMOS sensors.

Not only is the screen on your smart phone or TV likely inspected with image sensors manufactured at the Lake Avenue site, but image sensors manufactured at this facility are also on the surface of Mars, orbiting Jupiter and the Moon, and used in commercial satellites that monitor the Earth’s surface,” said Michael Miller, GM and director of operations at ON Semiconductor.

Manufacturing is a core competency for ON Semiconductor and the majority of ON Semiconductor’s manufacturing operations are done internally through the company’s industry leading cost structure,” said Bill Schromm, EVP and COO. “This expansion is important to our company, as it significantly increases our assembly capacity at the ON Semiconductor Rochester location.

Rochester is known for its innovations in digital imaging, including the design and development of state-of-the art image sensors over the past decades. Assembly and test has always been a key part of the equation, and as the resolution and complexity of the sensors continues to increase, these operations have become critical,” said Herb Erhardt, GM, Industrial Solutions Division. “The increased level of capability and capacity enabled by this expansion is our answer to meeting these critical market needs, and the fact that we are doing it here in Rochester speaks to the capabilities of the teams we build here.

CMOS Sensor Sales Grow at Record-Breaking Pace

IC Insights reports "The spread of digital camera applications in vehicles, machine vision, human recognition and security systems, as well as for more powerful camera phones will drive CMOS image sensor sales to an eighth straight record-high level this year with worldwide revenues growing 10% to $13.7 billion, following a 19% surge in 2017.

With the smartphone market maturing, sales growth in CMOS image sensors slowed to 6% in 2016, but strong demand in other imaging applications played a major factor in boosting revenues by 19% to $12.5 billion last year. Sales of CCD and other image sensor technologies fell 2% in 2017 to about $1.6 billion after rising 5% in 2016.

Overall, CMOS image sensors grabbed 89% of total image sensor sales in 2017 compared to 74% in 2012 and 54% in 2007. Unit shipments of CMOS imaging devices represented 81% of total image sensors sold in 2017 compared to 64% in 2012 and 63% in 2007.

In new smartphones, CMOS image sensors are also seeing a new wave of growth with the increase of dual-lens camera systems (using two sensors) for enhanced photography. Cellular camera phones accounted for 62% of CMOS image sensor sales in 2017, but that marketshare is forecast to slip to 45% in 2022. Automotive CMOS image sensors are projected to grow the fastest among major end-use applications through the five-year forecast, rising by a compound annual growth rate (CAGR) of 38.4% to about 15% of total CMOS image sensor sales in 2022 ($2.8 billion) while camera phone-generated revenues are expected to rise by a CAGR of just 2.2% to $8.6 billion that year.
"

Wednesday, May 09, 2018

TowerJazz and Newsight Imaging Announce CMOS Sensor for LiDAR

GlobeNewswire: TowerJazz and Newsight Imaging announced production of Newsight’s CIS chips and camera modules, customized for very high volume LiDAR and machine vision markets, combining sensors, digital algorithms and pixel array on the same chip.

Newsight’s image sensor chips feature cost effectiveness, low power consumption, high performance, and analog and digital integration. The NSI3000 sensor family, currently in mass production at TowerJazz’s Migdal Haemek, Israel facility, offers extremely high sensitivity pixels, enabling the replacement of expensive CCDs in many applications and is designed for programmable high frame rate speeds, allowing better analysis and reaction to events.

In addition, Newsight’s NSI5000, currently in development with TowerJazz at its fab in Israel, is an integrated LiDAR solution for long-range applications and includes a DSP controller which enables complex calculations for depth and machine vision. NSI5000 is used in 3D pulsed based LiDARs for automotive applications and is based on Newsight’s eTOF (enhanced time-of-flight), which bridges the gap between short-distance iTOF (indirect time-of-flight) and the long distance automotive requirement, by extending the DR while retaining high accuracy.

We chose TowerJazz for its advanced pixel technology, specially customized for our CMOS image sensor chips addressing very high volume markets. Together with our technology, we were able to demonstrate a 4X better sensitivity to our customers. TowerJazz’s CIS offering is proven in the industry and we are pleased to manufacture locally in Israel with a leader in the global analog foundry space,” said Eli Assoolin, CEO, Newsight Imaging.

With our high-end pixel offering, tailored to specific product and application needs, we are able to provide advanced technology used for high dynamic range CMOS sensors and solutions for the growing LiDAR and automotive markets. We are very happy to work closely with Newsight Imaging to provide market leading solutions and achieve quick time to market. They have shown to be an extremely fast-moving customer and we have a lot of confidence in their success,” said Avi Strum, TowerJazz SVP and GM, CMOS Image Sensor Business Unit.

Snapdragon 7XX Series to Support Triple Cameras

A number of on-line sources quote SuggestPhone site on the specs of the oncoming Qualcomm 700 series of mid-range Snapdragon processors. The add support for triple cameras, as well as AI processing accelerators:


DPReview reports that 2019 iPhone might have a triple camera too.

SensL Explains SiPM Principles

SensL publishes SiPM explaining video in its new capacity of a part of ON Semi Imaging group:

ON Semi Acquires SensL

BusinessWire: ON Semiconductor acquires SensL Technologies Ltd. SensL, based in Ireland, is specializing in Silicon Photomultipliers (SiPM), Single Photon Avalanche Diode (SPAD) and LiDAR sensing products for automotive, medical, industrial and consumer markets. ON Semiconductor intends to extend its market in automotive sensing applications for ADAS and autonomous driving with expanded capabilities in imaging, radar and LiDAR. By combining this acquisition in Ireland with previously acquired radar technology and design centers in Israel and United Kingdom, ON Semi aims to provide a comprehensive set of sensor solutions for next generation highly autonomous vehicles.

The entire SensL team, founded by Carl Jackson and led by Bryan Campbell, have done a great job in bringing SiPMs to market and we look forward to expanding their market success and continuing their product portfolio in LiDAR, medical imaging and radiation detection,” said Taner Ozcelik, SVP and GM, Image Sensor Group. “The automotive sensor fusion demand is growing at an accelerated pace with a need for additional sensor technologies that are provided by the SensL team. Expanding our sensor technology assets and design capacity in Ireland will allow us to extend our leadership in established segments and to deliver new, world-class solutions for emerging segments.

Tuesday, May 08, 2018

European Imaging & Sensing Summit

European Imaging & Sensing Summit to be held in Grenoble, France on Sept. 19-20, 2018 publishes its partial agenda:

  • Image and Vision Sensors, Systems and Applications for Smart Cities
    Thierry Ligozat, Chief innovation officer, Teledyne e2v
  • Ultra-high resolution global shutter CMOS image for demanding industrial applications
    Xinyang Wang, CEO, Gpixel
  • TBA
    Taner Ozcelik, Senior Vice President & General Manager, Image Sensor Group, ON Semiconductor
  • Ge-on Si optical sensors for normal incidence - A low-cost alternative to InGaAs?
    Martin Steglich, Research group leader, Friedrich Schiller University Jena
  • Custom Bayer filter approach for multispectral imaging
    Stéphane Tisserand, CEO, SILIOS Technologies
  • TBA
    Robert Bez, Senior Vice President of Technology, LFoundry
  • Light Field for Sensing
    Christian Perwass, Project Manager, Bosch

The Technology Showcase sessions are open to the companies of any size with innovative technologies and solutions:


Thanks JLJ for the pointer!

TowerJazz Q1 2018 Report

SeekingAlpha publishes TowerJazz Q1 2018 earnings call transcript with update on the foundry's image sensor business:

"In the CMOS image sensor business unit, where we see recently very good results of our smallest in the world 2.5 micron global shutter pixel developed in our 300-millimeter 65-nanometer Uozu fab in Japan. We have two lead customers in the machine basic market, we have received silicon and plant provide prototype soon to their anxiously-awaiting customers. One of the products is a high resolution sensor for machine vision applications that with this pixel size gives a 4x resolution improvement with industry best current and quantum efficiency yielding much better performance than what is otherwise available in the market. We are supplying a wide range of pixel sizes for different machine vision applications on global shutter technology platform that will enhance our leadership in this market. We expect to start production on this platform by the end of the year.

In the medical X-ray market, we just prototyped a one dye per wafer device. We are one of the world's leading X-ray sensor suppliers and plan to ramp the high volume by the end of this year with three different products, all targeted to be high runners. In addition, as previously discussed, our 300-millimeter development of a 21/21 centimeter X-ray device for tiled and non-tiled applications has lead customer having demonstrated excellent pixel performance on prototypes and as well as excellent product yield. These results have attracted activities with other market leaders.

We are moving according to our plans with the leading DSLR camera supplier and we will start soon as second even more ambitious project, which includes stacked backside illuminated wafers for this market. This will continue to position us as a leading foundry for high-end photography applications. All these exciting activities will bear fruits in the coming two to three years in a steady high volume, high margin production as the world-leading provider of CMOS sensors for high-end applications. In addition, we are working today on several exciting projects that will drive very high volume in the augmented reality and virtual reality markets for both sensors and displays, and on a very unique embedding of artificial intelligence into imaging and other sensors. This should yield very high volumes in the coming three to five years and beyond.

We continue to produce infrared sensors not only for iOS and Android mobile platforms as previously announced, but with strong presence across industrial and specifically growing and automotive night-vision driver systems application.
"

LiDAR Technologies Compared

ArsTechnica compares different LiDAR technologies in a form of interview with Ouster CEO and co-founder Angus Pacala. Ouster develops a spinning mechanical LiDAR similar to Velodyne. Angus also co-founded Quanergy, so that he was closely involved in solid-stage LiDAR design. Few quotes:

"Pacala pointed out a couple of big advantages of the classic spinning design. The most obvious one is the 360° field of view. You can stick one lidar unit on the top of a car and get a complete view of a car's surroundings. Solid state lidars, in contrast, are fixed in place and typically have a field of view of 120° or less. It takes at least four units to achieve comparable coverage with a solid-state sensor.

Another less obvious advantage, Pacala says, is that eye safety rules allow a moving laser source to emit at a higher power level than a stationary one. With a scanning solid-state unit, putting your eye inches from the laser scanner could cause 100 percent of the laser light to flood into the eye. But with a spinning sensor, the laser is only focused in any particular direction for a fraction of its 360° rotation. A spinning lidar unit can therefore put more power into each laser pulse without creating risk of eye damage.

The tiny mirrors in MEMS systems can only reflect so much light. That makes it inherently difficult to bounce a laser beam off a distant object and detect the return flash.

The phased-array approach tends to produce beams that diverge more than other techniques, making it hard to achieve a combination of long range, high scanning resolution, and wide field of view.

With flash lidar, the light from each flash is spread over the entire field of view, which means that only a fraction of the light strikes any particular point. And each pixel in the photodetector array is necessarily quite small, limiting the amount of returned light it can capture.

Overall, my conversation with Ouster's Pacala made me less bullish about improvements in lidar costs. Prices are falling, as illustrated by Velodyne's 50 percent price cut for its 16-laser unit this year. And if you're willing to settle for a lidar with lower range and resolution, you can find units that cost a few thousand or even a few hundred dollars.

But the best lidar units—and possibly the only ones that are suitable for fully driverless cars—still seem to cost tens of thousands of dollars.

The headline originally described Ouster's lidar as "bulky," but Pacala emailed to dispute that: "The Innoviz Pro and the proposed Innoviz One are both way bigger, so is AEye's iDAR, so is Continental/ASC's flash lidar, so is Princeton Lightwave's, so is Luminar's old and new device, so are all three of Cepton's products, and so is the Quanergy's S3.
"

Monday, May 07, 2018

China Face Recognition Network in Action

A BBC reporter has been given a demo of China's state-wide face recognition capabilities:

Microsoft Announces 3D Camera "Project Kinect for Azure"

PRNewswire: Microsoft announces its latest 3D ToF camera: "A new initiative, Project Kinect for Azure — a package of sensors from Microsoft that contains our unmatched time of flight depth camera, with onboard compute, in a small, power-efficient form factor — designed for AI on the Edge. Project Kinect for Azure brings together this leading hardware technology with Azure AI to empower developers with new scenarios for working with ambient intelligence."

Microsoft AR visionary and architect Alex Kipman reveals the new ToF camera spec in his LinkedIn post:

  • Highest number of pixels (megapixel resolution 1024x1024)
  • Highest Figure of Merit (highest modulation frequency and modulation contrast resulting in low power consumption with overall system power of 225-950mw)
  • Automatic per pixel gain selection enabling large dynamic range allowing near and far objects to be captured cleanly
  • Global shutter allowing for improved performance in sunlight
  • Multiphase depth calculation method enables robust accuracy even in the presence of chip, laser and power supply variation.
  • Low peak current operation even at high frequency lowers the cost of modules

Some of the claims are objectionable. For example, Odos Imaging used to sell a higher resolution 4MP ToF camera few years ago. The global shutter for better sunlight performance is somewhat unclear claim too.

"Earlier this year, Cyrus Bamji, an architect on our team, presented a well-received paper to the International Solid-State Circuits Conference (ISSCC) on our latest depth sensor. This is the sensor... that will give the next version of HoloLens new capabilities."


And some more Microsoft marketing:

"Microsoft announced Project Kinect for Azure, a package of sensors, including our next-generation depth camera, with onboard compute designed for AI on the Edge. Building on Kinect's legacy that has lived on through HoloLens, Project Kinect for Azure empowers new scenarios for developers working with ambient intelligence. Combining Microsoft's industry-defining Time of Flight sensor with additional sensors all in a small, power-efficient form factor, Project Kinect for Azure will leverage the richness of Azure AI to dramatically improve insights and operations. It can input fully articulated hand tracking and high-fidelity spatial mapping, enabling a new level of precision solutions."

A clip from keynote of the company CEO Satya Nadela at Microsoft Build:

Saturday, May 05, 2018

Two-Tap Pixel for Heart Rate Detection

VLSI Symposium 2018 publishes a preview of Shizuoka University, Brookman, and Chiba University paper "A Two-Tap NIR Lock-In Pixel CMOS Image Sensor with Background Light Cancelling Capability for Non-Contact Heart Rate Detection."

"Sensor technologies, whether for the Internet of Things, industrial electronics, or biomedical applications, have been and continue to be an important part of the VLSI Symposia. This year, both Technology and Circuits papers fall into this category. First, C. Cao from Shizuoka University will present a CMOS image sensor using two-tap near infrared lock-in pixels for non-contact heart rate detection. The two-tap pixels are used to cancel background light, achieving >98% detection precision even in the presence of sinusoidal varying bright ambient light, comparable to the latest visible-band-based ISP-assisted method. Fabricated in a 0.11um CIS technology, the achieved maximum modulation ratio is 90%, as well as a low random noise of 1.1e-rms."

Friday, May 04, 2018

Interview with Boyd Fowler, Omnivision CTO

3DInCities publishes an interview with Boyd Fowler, Omnivision CTO. Omnivision Nyxel NIR enhancing sensor has been a winner of 3D InCites Award in the Device of the Year category. Few quotes from the interview:

"...drivers of digital imaging technology have converged to two distinct paths: digital photography and machine-vision applications. The former has been the main driver for some time. The latter is a relatively new and growing market space.

Ten years down the road, everything you own could have a camera in it...

High reliability is mission critical for the automotive and medical markets, but the industry doesn’t always see it that way. We consider our automotive and medical image sensors almost as a separate business from our consumer segments to ensure their reliability.

...cost reduction has been another hurdle to overcome. “20 years ago, CMOS image sensors were boutique technologies and very expensive,” said Fowler. “While performance is increasing, the expectation is that the price will drop. Perpetuating that is an ongoing challenge.”

...today, the packaging used to meet reliability requirements ends up being larger and bulkier than other camera modules. What is needed is high reliability combined with very small packaging.
"

Yole on LiDAR Patents

Yole publishes "LiDAR for Automotive - Patent Landscape Analysis" tracing the first automotive LiDAR patents back to 1934: