Image Sensors World        Go to the original article...

KAIST research team led by Prof. Hoi-Jun Yoo developes CNNP chip (CNN Processor), that runs AI algorithms with ultra-low power, and K-Eye, a face recognition system using CNNP. The system was made in collaboration with a start-up company, UX Factory.

The “Always-on” image sensor, can determine if there is a face in its camera range. Then, it can capture frames and set the device to operate only when a face exists, reducing the standby power significantly. CNNP achieved 97% face recognition accuracy while consuming only 0.62mW of power. Both chips were developed by Kyeongryeol Bong, a PhD student under Prof. Yoo and presented at ISSCC in February.


A Youtube video demos the K-Eye operation:



Update: EurekaAlert also publishes a die photo:

Go to the original article...

Image Sensors World        Go to the original article...

TechInsights starts a new blog post series "A Survey of Enabling Technologies in Successful Consumer Digital Imaging Products" based on Ray Fontaine's presentation at IISW 2017. Part 1 talks about PDAF technologies and is very insightful:

Go to the original article...

Image Sensors World        Go to the original article...

BusinessWire: Technavio report on automatic production inspection camera market forecasts the market size to reach $1,436.46 million by 2021, at a CAGR of nearly 10%. The largest part of the market is UV cameras:

Go to the original article...

Image Sensors World        Go to the original article...

BusinessWire: Presto Engineering, an outsourced operations provider to semiconductor device manufacturers, now offers a turnkey industrialization service for image sensors.

Larger-volume CIS vendors typically keep their manufacturing operations in-house. However, a fragmented market at medium volumes, perhaps as large as $500M aggregated across all segments, provides significant opportunities for fabless CIS companies who can benefit greatly from outsourced test and characterization solutions.

Presto’s IP750Ex test solution provides a large (80x100mm), uniform (+/-2%), high-intensity illumination field at a range of temperatures. Its highly parallel test capability for complex devices (as large as 32M pixels at speeds up to 200MHz) enables fast ramps without compromising cost-of-test. Its extension into IR and its ability to test 12-, 8- and 6-inch wafers across the full automotive temperature range (-40°C to +150°C) qualifies it uniquely for automotive vision integrated circuits (ICs).

Philippe Rommeveaux, CEO of France-based Pyxalis states, “Pyxalis is very excited to use Presto Engineering and its state-of-the-art imaging platform to test custom image sensors. Local support and interaction make this solution very efficient.”

Presto Engineering provides outsourced operations from wafer procurement through final product shipment, and much more. The image sensor program and cleanroom facility are located in Meyreuil, France, near Aix-en-Provence.

Go to the original article...

Image Sensors World        Go to the original article...

Melexis new ToF chipset includes the MLX75023 1/3-inch ToF sensor and the MLX75123 companion IC that embeds many of the external components normally required to develop a 3D vision solution. With this high level of integration, designers no longer need external FPGAs and ADCs thereby reducing size, design cost, product cost and time-to-market.

The MLX75023 ToF sensor is said to offer the world’s smallest pixel at QVGA resolution with 63 dB linear DR and sunlight robustness, thanks to Melexis' advanced pixel technology (15um Softkinetic pixels, that is). The MLX75123 companion chip directly interfaces the sensor IC to a host MCU and provides rapid readout of data from the sensor.

Gaetan Koers, Product Manager for Melexis commented: “We are very excited about the possibilities the launch of this chipset unleashes. The simplicity that it brings to the design process will ensure that more applications will be able to benefit from 3D TOF vision. In this fast-moving technology sector the in-built future proofing will mean that our customers will be able to remain at the leading edge for years to come.”


Melexis opens a Youtube channel devoted to its new ToF platform:

Go to the original article...

Image Sensors World        Go to the original article...

ALA News: Sofradir designs its first supersize 4MP SWIR sensor with 15um pitch for ESA space observation programs. The ALFA 2Kx2K (Astronomy Large Focal Array) 15µm pitch SWIR detector is said to provide the highest levels of performance in QE, dark current and noise in order to optimize IR observations for astrophysicists. The SWIR detector is based on a compound semiconductor material, mercury cadmium telluride (HgCdTe)and includes a readout circuit (ROIC) with a source follower per detector (SFD) at the input stage.

Go to the original article...

Image Sensors World        Go to the original article...

Semiconductor Engineering publishes "CMOS Image Sensors (CIS): Past, Present & Future" article by Sofiane Guissi, staff engineer at Coventor. Among other historical comparisons and tables, there is one from www.3DIC.org talking about CIS stacking timeline, not including Sony 2017 3-layer CIS mass production milestone:

Go to the original article...

Image Sensors World        Go to the original article...

TI publishes a nice video showing a fusion between so different views from an automotive radar and a camera:

Go to the original article...

Image Sensors World        Go to the original article...

TI publishes a nice video showing a fusion between so different views from an automotive radar and a camera:

Go to the original article...

Image Sensors World        Go to the original article...

Omnivision kindly sent me few pictures showing how small the new 1/36-inch OV6948 image sensor is:

Go to the original article...

Image Sensors World        Go to the original article...

Dual fron and dual rear camera combos appear in more and more low-end camera phones in China. DeviceSpecifications reports that the new Oukitel U22 features 13MP+2MP rear cameras and 8MP+2MP front cameras.


Meanwhile, Doogee comes up with its second quad camera phone, model X30 (see the first one here). The phone features 5MP + 5MP on the front side and 8MP + 8MP on the rear (source: DeviceSpecifications):

Go to the original article...

Image Sensors World        Go to the original article...

EDN publishes a July 2016 video interview with Analog Devices representative showing SNAP image sensing technology:

Go to the original article...

Image Sensors World        Go to the original article...

BusinessWire: THine has started, from June 2017, mass-production and delivery of THCV241-Q MIPI extender which makes it possible to transmit image data up to 15m by converting MIPI CSI-2 interface to V-by-One HS interface. The 8MP 30fps image data can be transmitted by V-by-One HS with one pair of shielded twisted pair wire or one coaxial wire (or one shielded twisted quad cable). Such extenders are popular in automotive and security cameras.

Features of THCV241-Q:

• MIPI CSI-2 input (1/2/4lane, 1.2Gbps/lane), V-by-One HS output (1/2lane, 4Gbps/lane)
• Capable of transmitting image data of 8Mpixel (4K2K) 60fps at the maximum (in case of V-by-One HS 2 lane)
• Control signal bridge function of GPIO/UART/2 wire serial interface
• Error detection function including CRC (image data transmission path/control data transmission path)
• Function of distributing image data (in case of V-by-One® HS 2lane)
• Follow AEC-Q100 Grade 2

Go to the original article...

Image Sensors World        Go to the original article...

PRNewswire: OmniVision introduces the OV6948, one of the world's smallest commercially available image sensors. Initially developed for small-outer-diameter medical endoscopes and catheters, the OV6948 can be designed into a wide range of applications, including dental, veterinary, IoT, industrial, wearable and forensic devices. The OV6948 leverages OmniBSI+ pixel technology in a package that measures just 0.575 mm × 0.575 mm.

The product features:

• 1/36-inch optical format
• 200 × 200 pixel capture at 30 frames per second (FPS)
• Sensitivity: 1000 mV/lux-sec
• Analog data output with the ability to transmit up to 4 meters with minimal noise
• 0.575 mm x 0.575 mm package size (4 pads total) for easy integration

The OV6948 is currently sampling and is expected to start mass production in Q2 2018.

Go to the original article...

Image Sensors World        Go to the original article...

ALA: ULIS announces significant increase in its R&D spending for 2017 to reach €15M ($16.8M). The company generated €61.5M ($69.3M) in sales in 2016, a rise of nearly 30% compared to 2015. The new level of R&D investment will represent 20% of its annual sales. Previously, the company pledged 10% of its annual revenue to R&D.

Since its founding in 2002 as a spin-off from CEA-Leti, ULIS has become the second largest producer of microbolometers for defense, surveillance, thermography, firefighting, outdoor leisure and automotive markets. The company that started with 35 staff now employs 200. It has experienced double-digit growth, averaging 20% per year since its inception.

In addition, ULIS has injected €70M ($78.7M) to improve production tools and processes, enabling an increase in volume production from a couple of thousand units per year to several hundred thousand. It will soon deliver its one millionth component.

Go to the original article...

Image Sensors World        Go to the original article...

Sony patent application US20170148832 "Solid-state image pickup device and method of driving the same" by Kazuyoshi Yamashita reveals the company's dual CG BSI pixel layout:

Go to the original article...

Image Sensors World        Go to the original article...

FinancialTimes, Telegraph, South China Morning Post: Face recognition becomes quite widely used in China:

• Baidu has developed facial recognition gates for entry into its offices and ticketed attractions.
• Ant Financial, the online payments affiliate of Alibaba, lets its 450M users log in to their online wallets by taking a selfie
• China Construction Bank allows customers to pay with their faces at some vending machines
• Car-hailing service Didi Chuxing is using face recognition to verify drivers’ identities
• Facial recognition has been installed in ATM machines, KFC restaurants and even public toilets
• Baidu used facial recognition technology to help reunite lost children in China with their families

Beijing-based startup Face++ raised $100m in its third fundraising round in December 2016, which granted it unicorn status by a valuation of more than $1bn. Another 3-year old Beijing-based start-up, SenseTime, boasting 99% face recognition accuracy, is valued at more than $1bn too.

Go to the original article...

Image Sensors World        Go to the original article...

ISSCC publishes a number of old demo session videos. Some of them talk about image sensors:

"A 1920×1080 3.65μm-pixel 2D/3D image sensor with split and binning pixel structure in 0.11pm standard CMOS" by Samsung:



"A 1kPixel CMOS camera chip for 25fps real-time terahertz imaging applications" by University of Wuppertal, Germany:



"A global-shutter CMOS image sensor with readout speed of 1Tpixel/s burst and 780Mpixel/s continuous" by Tohoku University, Japan:

Go to the original article...

Image Sensors World        Go to the original article...

MediSens Conference publishes some its 2016 video presentations on-line:

Martin Wäny, CEO, Awaiba SA, Member of AMS group talks about "Miniature camera modules for visualization in micro invasive surgery:"



Koichi Mizobuchi, Deputy General Manager, Medical Imaging Technology Department, Olympus talks about "Medical endoscopic imaging and requirements for future image sensor design:"



Renato Turchetta, CEO, Wegapixel, now CEO at IMASENIC, presents "How CMOS can be further leveraged to advance medical imaging:"



Biju Jacob, Senior Engineer, X-ray Detector Development, General Electric presents "Advances in CMOS Wafer scale imagers for medical imaging:"



There are many more presentations at MediSens Youtube channel.

Go to the original article...

Image Sensors World        Go to the original article...

Tractica publishes an interview with Allan Benchetrit, CEO of Algolux, talking about the differences between ISP pipeline optimizations for a vision processing and a human eye:

"A CV algorithm looks for different things than the human eye, so manual tuning and visual inspection can never achieve the best possible result. The ISP settings that deliver the most pleasing visual result (denoised, etc.) actually remove valuable pixel data from the image or video stream, giving the classifier or object detector less to work with (see Figure 2 below)."

Go to the original article...

Image Sensors World        Go to the original article...

The ITE Transactions on Media Technology and Applications (MTA) announces a Special Section on “Advanced Image Sensor Technology” to be published in July 2018. The objective of the special section is focus on the advances in image sensor technologies. ITE MTA solicits the original research papers of the topics which include the following:

• CMOS image sensors, CCD image sensors, and related technologies
• Application specific image sensors
• Novel architecture, operation and materials for imaging devices
• Invisible imaging devices (IR, UV, X-ray, high-energy particles, ions, etc.)
• Machine vision
• Others

Papers must be submitted by November 15, 2017.

The two previous Special Sections on image sensors can be found here:
https://www.jstage.jst.go.jp/browse/mta/4/2/_contents
https://www.jstage.jst.go.jp/browse/mta/2/2/_contents

Go to the original article...

Image Sensors World        Go to the original article...

Albert Theuwissen announces the lecturer name for his Harvest Imaging Forum "Low-Noise Analog CMOS Circuit Design: from devices to circuits" to be held in December 2017. The lecturer is Christian Enz from EPFL, Switzerland.

Go to the original article...

Image Sensors World        Go to the original article...

EETimes: Corephotonics shares its experience on combining the images from two cameras - a wide angle and a tele- ones. The image fusion presents a number of challenges:

Image Fusion Must Deal with Dual Camera Occlusions

Wide and Tele Images with Different Depth of Field

Smart Decision Making

Image fusion artifacts in Apple's iPhone7+

Go to the original article...

Image Sensors World        Go to the original article...

BusinessWire: ON Semi new 29MP KAI-29052 CCD provides up to twice the sensitivity of the existing KAI-29050 in the wavelength range of 500 nm to 1050 nm. This enhanced pixel design retains isolation of charge from one photodiode to another, enabling this increase in sensitivity without any reduction to image sharpness (MTF). In addition, an improved amplifier design reduces read noise by 15%, increasing the linear DR to 66 dB.

“The evolving needs of industrial imaging applications such as high end security, machine vision, and aerial surveillance and mapping require continued advancements in the portfolio of image sensors serving this market,” said Herb Erhardt, VP and GM, Industrial Solutions Division, Image Sensor Group at ON Semiconductor. “The enhanced performance available from the KAI-29052 is a further demonstration of ON Semiconductor’s ongoing commitment to provide the most advanced CCD and CMOS image sensors for these demanding markets.”

Go to the original article...

Image Sensors World        Go to the original article...

CMOSIS has been renamed to ams Sensors Belgium.

Go to the original article...

Image Sensors World        Go to the original article...

Sony subsidiary Softkinetic presents a BSI ToF sensor with a 10µm pixel pitch, which the company calls "the industry's smallest". As as side note, Microsoft Kinect 2 camera has 10um ToF pixel too, available on the mass market since 2013. The Softkinetic's proprietary Current Assisted Photonic Demodulator ("CAPD") pixel architecture is capable of 100MHz high-speed modulation using drift current in the pixel, allowing for 1.5x longer distance. The new pixel light collection efficiency is equivalent to that of a 15µm pixel pitch FSI sensor, but using a BSI with a 10µm pixel pitch.

Go to the original article...

Image Sensors World        Go to the original article...

GPixel announces a 43MP global shutter sensor, a high QE BSI sensor and a line scan TDI sensor:


43MP GMAX0806 GS sensor features:

• 7915 × 5436 resolution
• pixel size 2.8μm
• maximum frame rate up to 180fps
• read the noise 2e-
• aimed to high-end industrial testing, Film machine, aviation railway detection, astronomical applications and more

GSENSE2020BSI features:

• 4MP resolution
• pixel size 6.5μm
• peak QE greater than 90%
• readout noise 0.8e-
• aimed to biological microscopy, spectral Imaging, astronomical imaging and other areas.

GL0816 linescan TDI sensor features:

• 8322 × 16 linear array
• on-chip 2 TDI
• maximum frame rate up to 130kHz
• peak QE higher than 70%
• read the noise 2.4e-
• aimed to high-speed industrial testing, machine vision

Go to the original article...

Image Sensors World        Go to the original article...

Bart Dierickx, Caeleste CTO and CEO, kindly permitted me to reproduce his IISW 2017 poster "How to hand-calculate MTF in front-side and backside illuminated image sensors" written by himself, Jean Bourgain, and Bert Luyssaert. Everybody is invited to download and use the spreadsheet with the quick calculations. Bart says he would be happy if as many people as possible review it, comment, and, possibly, find ways to improve it:

Go to the original article...

Image Sensors World        Go to the original article...

Albert Theuwissen concludes his IISW review series. Part 4 talks about Invisage sensor performance, GS pixels, and more.

Go to the original article...

Image Sensors World        Go to the original article...

International Image Sensor Society publishes part of the papers presented at 3rd International Workshop on Image Sensor and Systems (IWISS2016). The following papers have been published:

• [Invited-01] Ion implantation technology for image sensors (Prof. Nobukazu Teranishi, Dr. Genshu Fuse*, and Dr. Michiro Sugitani*, Shizuoka Univ./ Univ. of Hyogo, *Sumitomo Heavy Industries Ion Technology Co., Ltd., Japan)
• [Invited-02] 3D-stacking architecture for low-noise high-speed image sensors (Prof. Shoji Kawahito, Shizuoka Univ., Japan)
• [Invited-03] A Dead-time free global shutter stacked CMOS image sensor with in-pixel LOFIC and ADC using pixel-wise connections (Prof. Rihito Kuroda, Mr. Hidetake Sugo, Mr. Shunichi Wakashima, and Prof. Shigetoshi Sugawa, Tohoku Univ., Japan)
• [Invited-04] 3D stacked image sensor featuring low noise inductive coupling channels (Prof. Masayuki Ikebe, Dr. Daisuke Uchida, Dr. Yasuhiro Take*, Prof. Tetsuya Asai, Prof. Tadahiro Kuroda*, and Prof. Masato Motomura, Hokkaido Univ., *Keio Univ., Japan)
• [Invited-05] Low-noise CMOS image sensors towards single-photon detection (Prof. Min-Woong Seo, Prof. Keiichiro Kagawa, Prof. Keita Yasutomi, and Prof. Shoji Kawahito, Shizuoka Univ., Japan)
• [Invited-06] Bioluminescence imaging in living animals (Prof. Takahiro Kuchimaru, Prof. Tetsuya Kadonosono, Prof. Shinae Kizaka-Kondoh, Tokyo Inst. of Tech., Japan)
• [Invited-07] Restoration of a Poissonian-Gaussian color moving-image sequence (Prof. Takahiro Saito and Mr. Takashi Komatsu, Kanagawa Univ., Japan)
• [Invited-10] Power-efficient Delta Readout Scheme utilizing Multi-Column-Parallel SAR ADCs (Mr. Hyun-June Kim, Mr. Sun-Il Hwang, and Prof. Seung-Tak Ryu, KAIST, Korea)
• [Invited-12] Various ultra-high-speed imaging and applications by streak camera (Mr. Koro Uchiyama, Hamamatsu Photonics K. K., Japan)
• [Invited-13] Clock management in high range resolution time-of-flight range imagers (Prof. Keita Yasutomi and Prof. Shoji Kawahito, Shizuoka Univ., Japan)
• [Invited-15] Application of light-sheet microscopy to cell and development biology (Prof. Shigenori Nonaka, National Inst. for Basic Biology, Japan)
• [Invited-16] Non-contact video based estimation for heart rate variability spectrogram using ambient light by extracting hemoglobin information (Prof. Norimichi Tsumura, Chiba Univ. , Japan)
• [Invited-18] Extremely compact hyperspectral camera for drone and smartphone (Prof. Ichiro Ishimaru, Kagawa Univ., Japan)

Go to the original article...