Friday, June 22, 2018

Pico Presents Zense ToF Camera

NotebookItalia: China-based Pico presents Zense camera, its first foray into ToF sensing. The depth processing is based on Rockchip RV1108 with CEVA XM4 vision IP core:

Yole Forecast on 3D Sensing Market

Yole Developpement publishes "3D Imaging & Sensing 2018" report:

"Apple set the standard for technology and use-case for 3D sensing in consumer. From our initial depiction of the market in March 2017, the main gap is in illumination ASP, which is greater than expected. High expenses in dot and flood illumination VCSELs from Lumentum/II-VI/Finisar, along with the dot illuminator optical assembly from ams, are the biggest technology surprises powering Apple’s $1,000 [iPhone X] smartphone.

Yole Développement (Yole) expects the global 3D imaging & sensing market to expand from $2.1B in 2017 to $18.5B in 2023, at a 44% CAGR. Along with consumer, automotive, industrial, and other high-end markets will also experience a double-digit growth pattern.

The transition from imaging to sensing is happening before our eyes. Despite half-successful attempts like Xbox’s Kinect technology and Leap-Motion hand controllers, 3D sensing is now tracking towards ubiquitousness.

Oppo made the first announcement beginning of the year with Orbeec, while Xiaomi released the Mi 8 explorer edition with Mantis as a technology partner. We expect Huawei to release its own solution soon, probably partnering with ams and Sunny Optical.

...players like Himax are currently paying the price for a lower-performance offering and are struggling to get design-ins beyond AR/VR headsets for Microsoft.

Unlike previous sensing components, the responsibility of system design does not fall on the OEM - instead, a specialist is required, such as the Primesense team that Apple acquired in 2013, or other firms like Mantis, Orbbec, and ams, which want play the “specialist” role in the new 3D imaging & sensing ecosystem. Such players orchestrate the final solution while allowing room for the best in each sub-component category.

Is 3D imaging & sensing now ripe for disruption? Yole expects it will take at least 2 -3 years before any new solution start dramatically lowering total system cost.

Thursday, June 21, 2018

Smithers Report on Technology Challenges for Image Sensors in ADAS & Autonomous Vehicle

Smithers Apex publishes a report "Technology Challenges for Image Sensors in ADAS & Autonomous Vehicle Markets to 2023." Few quotes:

" vehicles approach full autonomy, the need for human-friendly colour reproduction becomes redundant, and the quality requirements on the image sensing for computer vision will increase further. A high dynamic range of 140 dB, resolutions of 8 MP for mono vision and at least 2MP for stereo vision cameras, frame rates beyond 60 fps, a sensitivity of several thousand electrons per lux and second are requested.

At the same time the noise performance should ensure a signal to noise ratio of 1 at 1 millilux, and an exposure time of less than 1/30 s. High dynamic range shall be achieved by as few as possible exposures to minimise motion blur. The compensation of the flickering of LED traffic signs and lights as described in the previous section is even more important for level 4 and 5 vehicles.

SWIR Startup Trieye Raises $3m of Seed Money

Globes: Israeli startup TriEye announces the completion of a $3m seed round led by Grove Ventures. TriEye develops SWIR sensors providing autonomous cars heightened visual capabilities in restricted visual conditions at significantly reduced cost. The technology is based on many years of research at the Hebrew University of Jerusalem by Prof. Uriel Levy.

TriEye cofounder and CEO Avi Bakal says, "The company is offering capabilities that in the past were only accessible to defense and aerospace industries and at a minimal cost compared with the past. This capability improves the safety of advanced driver-assistance systems and is a major step forward to the extensive adoption of the technology by the carmakers."

AutoSens Brussels Agenda

Autosens Brussels to be held in September 2018 announces its preliminary speakers list. There is quite a lot of image sensors presentations:
  • Introduction to the world of Time-of-Flight for 3D imaging, Albert Theuwissen, Harvest Imaging (Tutorial)
  • Do we have a lidar bubble? Panel discussion
  • Vehicle perception of humans – what level of image quality is needed to recognise behavioural intentions, Panel discussion
  • A review of the latest research in photonics-sensor technologies in automotive, Michael Watts, Analog Photonics
  • Automotive HDR imaging – the history and future, Mario Heid, Omnivision
  • Self-driving cars and lidar, Simon Verghese, Waymo
  • Tier 1 achievements with solid state lidar, Filip Geuens, Xenomatix
  • An Approach to realize “Safety Cocoon”, Yuichi Motohashi, Sony
  • How AI/Computer Vision affects camera design and SOC design, Marco Jacobs, videantis
  • Objective and Application Oriented Characterisation of Image Sensors with EMVA’s 1288 Standard, Bernd Jaehne, EMVA 1288 Chair
  • Review of IEEE P2020 developments, Patrick Denny, Valeo

Xintec Unable to Fill 12-inch WLP Line

Digitimes reports that Xintec has hard time recovering from the loss of its major investor and customer Omnivision. The company has decided to suspend its 12-inch WLP production line for a year due to disappointing demand for mass-market applications. The 12-inch line workforce will be transferred to 8-inch lines to improve revenue and profit.

Wednesday, June 20, 2018

Tunable Plasmonic Filters Enable Time-Sequential Color Imaging

ACS Photonics paper "Tunable Multispectral Color Sensor with Plasmonic Reflector" by Vladislav Jovanov, Helmut Stiebig, and Dietmar Knipp from Jacobs University Bremen and Institute of Photovoltaics Jülich, Germany proposes plasmonic reflectors for color imaging:

"Vertically integrated color sensors with plasmonic reflectors are realized. The complete color information is detected at each color pixel of the sensor array without using optical filters. The spectral responsivity of the sensor is tuned by the applied electric bias and the design of the plasmonic reflector. By introducing an interlayer between the lossy metal back reflector and the sensor, the reflectivity can be modified over a wide spectral range. The detection principle is demonstrated for a silicon thin film detector prepared on a textured silver back reflector. The sensor can be used for RGB color detection replacing conventional color sensors with optical filters. Combining detectors with different spectral reflectivity of the back reflector allows for the realization of multispectral color sensors covering the visible and the near-infrared spectral range.

To our knowledge for the first time a sensor is presented that combines a spatial color multiplexing scheme (side-by-side arrangement of the individual color channels) used by conventional color sensors with the time multiplexing scheme (sequential read-out of colors) of a vertically integrated sensor.

Daqri on Importance of Latency in AR/VR Imaging

Daqri Chief Scientist Daniel Wagner publishes an article "Motion to Photon Latency in AR and VR applications:"

Daqri tells about its latency optimization achievements: "Overall, what is the final latency we can achieve with such a system? The answer is: It depends. If we consider rendering using the latest 6DOF pose estimates as the last step in our pipeline that produces fully correct augmentations then achieving a latency of ~17ms is our best-case scenario.

However, late warping can correct most noticeable artefacts, so it makes sense to treat it as valid part of the pipeline. Without using forward prediction for late warping, we can achieve a physical latency of 4 milliseconds per color channel in our example system. However, 4 milliseconds is such a short length of time, making forward prediction almost perfect and pushing perceived latency towards zero. Further, we could actually predict beyond 4ms into the future to achieve a negative perceived latency. However, negative latency is just as unpleasant as positive latency, so this would not make sense for our scenario.

Velodyne Lidar 101

Velodyne publishes a promotional video explaining its technology:

Tuesday, June 19, 2018

Lucid Demos Sony Polarization Sensor

LUCID Vision Labs demos Sony IMX250MYR polarized color sensor in its Phoenix camera family. The 5 MP GS sensor with 3.45µm pixel and frame rates of up to 24 fps is based on the popular IMX250 Sony Pregius CMOS color sensor with polarizing filters added to the pixel. The sensor has four different directional polarizing filters (0°, 90°, 45°, and 135°) on every four pixels:

Panasonic Develops 250m-Range APD-based ToF Sensor

Panasonic has developed a TOF image sensorthat uses avalanche PD (APD) pixels and is capable of capturing range imaging of objects up to 250m even at night with poor visibility (there is no mention what is the range in mid-day sunlight). The new sensor applications include automotive range imaging and wide-area surveillance in the dark.

The ToF pixel includes an APD and an in-pixel circuit that integrates weak input signals to enables the 3D range imaging 250 m ahead. The sensor resolution is said to be the world's highest 250,000 pixels for a sensor based on electron-multiplying pixels. This high integration is achieved through the lamination of the electron multiplier and the electron storage as well as the area reduction of APD pixels.

The key innovative technologies are:
  • The area of APD pixels is significantly reduced while the multiplication performance is maintained through the lamination of the multiplier that amplifies photoelectrons and the electron storage that retains electrons.
    The APD multiplication factor is 10,000.
  • Long-range measurement imaging technology

Trioptics Active Alignment, Assembly and Testing of Camera Modules

Germany-based Trioptics demos Procam, its modular manufacturing line for active alignment, assembly and testing of camera modules in mass production:

Fiat-Chrysler Autonomous Car Relies on 5 LiDARs and 8 Cameras

Fiat-Chrysler 5-year plan presentation shows its Level 4 autonomous car with 5 LiDARs and 8 cameras. Most of the LiDARs are defined as "mid-range" possibly meaning their range is shorter than 200m for a cheaper price:

Monday, June 18, 2018

Gil Amelio on Patent Infrigements

Investors Business Daily publishes Gil Amelio article with a story of Pictos vs Samsung lawsuit:

"A typical small inventive company, Pictos Technologies, was put out of business after Samsung aggressively infringed its intellectual property.

Pictos invented an inexpensive image sensor that could be used in countless applications such as mobile phones and automobile cameras, to name only two. This next-generation Image Sensor was a follow-on to my dozen or so image-sensing patents that helped launch the solid-state image-sensor business years earlier. The Pictos technology, developed after years of investment and design, was protected by a portfolio of patents obtained at substantial cost.

In 2014, Pictos sued Samsung in federal court, alleging that it had "willfully infringed" its intellectual property. After years of costly litigation, the case went to trial, where Pictos lawyers introduced evidence that proved Samsung began as a Pictos customer, secretly copied its engineering designs and production process, and replicated them in Korea. Using our technology and its sizable scale, it went on to dominate this sector of the world electronics market.

Following lengthy litigation, the jury ruled in our favor and awarded substantial damages. The judge then trebled the damages based on "evidence of (Samsung's) conduct at the time of the accused infringement." Please note: Samsung's behavior was so egregious that the judge tripled the jury determination of the infringement costs to us.

That was just the first round, though. The verdict can be overturned on appeal, which, of course, Samsung has filed.

Update: Once we are at historical stuff, SemiWiki publishes Mentor Graphics CEO Wally Rhines memories from the early days of CCD and DRAM imagers in Stanford University in 1960s.

Microsoft Opens Access to Hololens Cameras

Microsoft opens access to a raw video stream from cameras in its Hololens AR headset, including 3D ToF camera:

"The depth camera uses active infrared (IR) illumination to determine depth through time-of-flight. The camera can operate in two modes. The first mode enables high-frequency (30 FPS) near-depth sensing, commonly used for hand tracking, while the other is used for lower-frequency (1-5 FPS) far-depth sensing, currently used by spatial mapping. In addition to depth, this camera also delivers actively illuminated IR images that can be valuable in their own right because they are illuminated from the HoloLens and reasonably unaffected by ambient light."

Sunday, June 17, 2018

Canon Explores Large Image Sensor Future

Canon publishes an article on its image sensor projects for academic and industrial customers.

The world's largest high-Sensitivity CMOS sensor is measuring ~20 cm square. As such, a 20-cm-square sensor is the largest size that can be manufactured on 300mm wafer, and is equivalent to nearly 40 times the size of a 35 mm full-frame CMOS sensor:

Canon has spent many years working to reduce the pixel size for CMOS sensors, making possible a pixel size of 2.2 µm for a total of approximately 120MP on a single sensor. The APS-H size (approx. 29 x 20 mm) CMOS sensor boasts approximately 7.5 times the number of pixels and 2.6 times the resolution of sensors of the same size featured in existing products. This sensor offers potential for a range of industrial applications, including cameras for shooting images for large-format poster prints, cameras for the image inspection of precision parts, aerospace cameras, and omnidirectional vision cameras.

Friday, June 15, 2018

TrinamiX Paper in Nature

Nature publishes BASF spin-off TrinamiX paper "Focus-Induced Photoresponse: a novel way to measure distances with photodetectors" by Oili Pekkola, Christoph Lungenschmied, Peter Fejes, Anke Handreck, Wilfried Hermes, Stephan Irle, Christian Lennartz, Christian Schildknecht, Peter Schillen, Patrick Schindler, Robert Send, Sebastian Valouch, Erwin Thiel, and Ingmar Bruder.

"We present the Focus-Induced Photoresponse (FIP) technique, a novel approach to optical distance measurement. It takes advantage of a universally-observed phenomenon in photodetector devices, an irradiance-dependent responsivity. This means that the output from a sensor is not only dependent on the total flux of incident photons, but also on the size of the area in which they fall. If probe light from an object is cast on the detector through a lens, the sensor response depends on how far in or out of focus the object is. We call this the FIP effect. Here we demonstrate how to use the FIP effect to measure the distance to that object. We show that the FIP technique works with different sensor types and materials, as well as visible and near infrared light. The FIP technique operates on a working principle, which is fundamentally different from all established distance measurement methods and hence offers a way to overcome some of their limitations. FIP enables fast optical distance measurements with a simple single-pixel detector layout and minimal computational power. It allows for measurements that are robust to ambient light even outside the wavelength range accessible with silicon.

In this paper, we demonstrated the measurement principle at distances up to 2 m and showed a resolution of below 500 µm at a distance of 50 cm. In the Supplementary Information S7, distance measurements up to 70 m can be found.

F-35 Gets 6 Cameras for Surround View

PRNewswire: Surround view cameras reach defense industry. Lockheed Martin selectes Raytheon to develop and deliver the next generation Distributed Aperture System (DAS) for the F-35 fighter jet.

The F-35's DAS collects and sends high resolution, real-time imagery to the pilot's helmet from six IR cameras mounted around the aircraft, allowing pilots to see the environment around them – day or night. With the ability to detect and track threats from any angle, the F-35 DAS gives pilots situational awareness of the battlespace.

imec is Back to Film, Organic Film

imec promotes its organic film image sensors:

"We demonstrated a first film measuring 6 by 8 cm – which can check 4 fingers simultaneously – and which has a resolution of 200ppi. The second film – designed for a single fingerprint – has a resolution of 500ppi. This level of accuracy is what would be typical for the FBI to identify someone correctly.

The image sensors detect visible light between 400 and 700 nm that is reflected by the skin. They can also detect light that penetrates the skin before being reflected. This latter feature is of value for detecting a heartbeat, which provides an extra security check.

The fingerprint and palm print sensor is made up of a layer of oxide thin-film transistors with organic photodiodes on top. These photodiodes can then be ‘tuned’ by using a different organic material so that they detect a different wavelength, such as near infrared. This enables the vein pattern in a hand to be visualized, which is even more precise for accurate identification than a palm print.

In addition to this fingerprint scanner based on photodiodes and light, imec and Holst Centre are also working on a scanner that uses thermal sensors (PYCSEL project). Once again a lower layer of oxide thin-film transistors is used. The upper layer is a material that measures electric temperature changes. The fingerprint is then detected indirectly by local variations in temperature changes that correspond with the pattern of the fingerprint. Here again a resolution of 500ppi is achievable.

Thursday, June 14, 2018

ON Semi Talks about Automotive Pixel Technologies

AutoSens publishes an interview with ON Semi talking about "Super Exposing" pixel that reduces LED flicker and other ON innovations for the automotive market:

Mazda CX-3 SUV Features Nighttime Pedestrian Detection

Nikkei: Mazda CX-3 compact SUV comes with, as a standard feature, an automatic emergency braking system that supports nighttime pedestrians detection:

"Nighttime pedestrians are detected by the monocular camera. To support nighttime pedestrians, in terms of software, the logic of detecting pedestrians was improved, enhancing the accuracy of recognizing pedestrians at night. Its hardware was also improved to increase the speed of exchanging data between the [Mobileye] EyeQ3 image processing chip and memory."

Wednesday, June 13, 2018

More AutoSens Detroit Interviews

AutoSens publishes more interviews from Detroit:

Xenomatix talks about many design wins for its LiDAR:

Tetravue talks about its technology:

FLIR talks about thermal camera for automotive applications:

Algolux talks about its ML algorithms:

3D Imaging with PDAF Pixels

OSA Optics Express publishes a paper "Depth extraction with offset pixels" by W. J. Yun, Y. G. Kim, Y. M. Lee, J. Y. Lim, H. J. Kim, M. U. K. Khan, S. Chang, H. S. Park, and C. M. Kyung, KAIST, QiSens, and Kongju National University, Korea.

"Numerous depth extraction techniques have been proposed in the past. However, the utility of these techniques is limited as they typically require multiple imaging units, bulky platforms for computation, cannot achieve high speed and are computationally expensive. To counter the above challenges, a sensor with Offset Pixel Apertures (OPA) has been recently proposed. However, a working system for depth extraction with the OPA sensor has not been discussed. In this paper, we propose the first such system for depth extraction using the OPA sensor. We also propose a dedicated hardware implementation for the proposed system, named as the Depth Map Processor (DMP). The DMP can provide depth at 30 frames per second at 1920 × 1080 resolution with 31 disparity levels. Furthermore, the proposed DMP has low power consumption as for the aforementioned speed and resolution it only requires 290.76 mW. The proposed system makes it an ideal choice for depth extraction systems in constrained environments."

Tuesday, June 12, 2018

NIT Demos Log Sensor with LED Flicker Suppression

New Imaging Technologies publishes a demo of its NSC1701 sensor featuring LED flicker suppression mode:

Depth Sensing: From Exotic to Ubiquitous

Embedded Vision Alliance publishes a video lecture "The Evolution of Depth Sensing: From Exotic to Ubiquitous" delivered by Erik Klaas, CTO of 8tree in September 2017.

ams 48MP 30fps Full-Frame GS Imager Enters Mass Production

BusinessWire: ams announces its CMV50000, a high-speed 48MP global shutter CMOS sensor for machine vision applications, has entered into mass production and is available for purchase in high volumes now.

The CMV50000, which features a 35mm-format 7920 x 6004 array of 4.6µm-sized pixels based on a 8T pixel architecture, operates at 30 fps with 12-bit pixel depth at full resolution or a binned 4K and 8K modes, and even faster – up to 60 fps – with pixel sub-sampling at 4K resolution.

The sensor offers 64dB optical DR at full resolution and up to 68dB in binned 4K mode. The image sensor benefits from the implementation of sophisticated new on-chip noise-reduction circuitry such as black-level clamping, enabling it to capture high-quality images in low-light conditions.

The superior imaging performance of the CMV50000 was recognized earlier in 2018 when it was named the Biggest Breakthrough Development at the Image Sensors Europe Awards 2018.

During recent months, ams has seen great demand for the CMV50000 from design teams developing new automated optical inspection systems and vision systems for testing flat panel displays,” said Wim Wuyts, Marketing Director for Image Sensors at ams. “The CMV50000 is now fully qualified and available to these manufacturers in production volumes. It is also about to be supported by a full demonstration system for evaluating the sensor’s performance.

Both the monochrome and color versions of the CMV50000 are available in production volumes now. The per unit pricing is €3,450.

ams investors presentation dated by December 2017 details the company's strategy in imaging:

ams Compares 3D Imaging Approaches

ams investors presentation from Dec 2017 compares different 3D imaging technologies:

Monday, June 11, 2018

AutoSens Detroit Inteviews

AutoSens publishes a number of videos from its Detroit conference:

Sony Depth Sensing Demo:

SensL talks about its acquisition by ON Semi:

Omnivision Announces New Sensors for Medical Applications

OmniVision announces the 1/11-inch OH01A HD sensor employing the 1.1um PureCel-S pixel stacked-die architecture to provide the compact size, high resolution and cost effectiveness required for the next generation of disposable and reusable endoscopes and catheters. The OH01A is the world’s first medical image sensor to provide 1280 x 800 resolution at a rate of 60 fps in a tiny 2.5 x 1.5 mm package.

As endoscopes reach further into the human body for diagnostics, surgeons require smaller image sensors with higher resolution and excellent image quality,” said Tehzeeb Gunja, principal marketing and business development manager at OmniVision. “At the same time, they need a wide viewing angle and close focus distance with low power consumption to reduce heat and improve patient comfort. The OH01A image sensor fulfills all these requirements.

Support for both MIPI and sub-LVDS output interfaces allows the OH01A to transmit image data over long distances. It also integrates one-time-programmable (OTP) memory to store manufacturing and calibration information.

The OH01A can also be autoclaved for reusable devices and sterilized for disposable ones. Samples of the OH01A are available now, and it is expected to enter volume production in Q4 2018.

OmniVision announces the 1/6-inch OV2741, a 1.4um PureCel-S pixel stacked-die sensor that extends medical image sensor family. The OV2741 provides 1080p at a rate of 60 fps and has a broadband double anti-reflective coating on its cover glass to eliminate glare, ghosting and reflections due to strong illumination.

The OV2741 is aimed to endoscopic devices used in diagnostic and surgical procedures, including airway-management (esophagoscopes, laryngoscopes, thorascopes, pleuroscopes, bronchoscopes, mediastinoscopes) and gastro-intestinal (gastroscopes, duodenoscopes, amniscopes) applications.

The OV2741’s low power consumption of just 90 mW, while operating at 60 fps and full HD resolution, keeps the distal tip of the endoscope cooler for greater patient comfort. Additionally, the OV2741 can be both autoclaved for reusable applications and sterilized for disposable ones.

Samples of the OV2741 are available now, and it is expected to enter volume production in Q3 2018.

OmniVision announces the OVMed, a mixed-signal ISP for medical, veterinarian and industrial endoscopy applications. The ISP comes in ASIC and FPGA versions.