Wednesday, October 26, 2016

SK Hynix to Try Foundry/Custom Model for its CIS Business

ETNews: As Hynix moves into 13MP sensors mass production at its 300mm M10 fab in Icheon in 2017, the company plans to reduce production of the low-priced low resolution CIS at 200mm M8 plant in Cheongju. Instead, it is going to outsource the low cost sensor production to other foundries such as DDIC and PMIC. Eventually, SK Hynix is going to stop low-resolution sensor production at M8 fab.

As a part of this plan, SK Hynix publicly announced that it has recently received all SiliconFile's assets of $3.98 million (4.5 billion KRW) from SiliconFile, which is 100% CIS design subsidiary. Silicon File is becoming SK Hynix’s CIS design house. SiliconFile is supposed to find new fabless customers. SK Hynix appointed Director Lee Dong-jae serving as the department head of foundry business department, as SiliconFile board director.

Receiving company assets from Silicon File and changing Silicon File into a design house indicate that SK Hynix is officially promoting its non-memory semiconductor business,” says ETNews industry source.

Hynix VP KD Yoo who has established and led Hynix image sensor business over the years, left the company and now is a Professor at Hanyang University.

SK Hynix in Cheongju. White building at 2 o’clock is M8

DENSO Works with Toshiba and Sony on ADAS

JCN Newswire: DENSO and Toshiba have reached a basic agreement to jointly develop a Deep Neural Network-Intellectual Property (DNN-IP), which will be used in image recognition systems which have been independently developed by the two companies for ADAS and automated driving technologies.

Because of the rapid progress in DNN technology, the two companies plan to make the technology flexibly extendable to various network configurations. They will also make the technology able to be implemented on in-vehicle processors that are smaller, consume less power, and feature other optimizations.

DENSO has been developing DNN-IP for in-vehicle applications. By incorporating DNN-IP in in-vehicle cameras, DENSO will develop high-performance, ADAS and automated driving systems. Toshiba will partition this jointly developed DNN-IP technology into dedicated hardware components and implement them on its in-vehicle image recognition processors to process images using less power than image processing systems with DSPs or GPUs.

DENSO also invests in the US-based machine learning startup THINCI. “We are thrilled DENSO is our lead investor,” said THINCI CEO Dinakar Munagala. “The automotive industry is one of the earliest adopters of vision processing and deep learning technology. DENSO’s investment in THINCI’s trailblazing solution confirms our own belief that our innovation has much to offer, not only in the automobile but in the wide range of everyday products.

JCN Newswire: DENSO announces that the image sensors provided by Sony have helped DENSO improve the performance of its in-vehicle vision sensors and can now detect pedestrians during night conditions.

Sony image sensors, which are also used in surveillance and other monitoring devices, enable cameras to take clear images of objects even at night. DENSO has improved the quality of Sony's image sensors in terms of ease of installation, heat resistance, vibration resistance, etc. to be used in vehicle-mounted vision sensors. DENSO has also used Sony's ISPs for noise reduction and optimization of camera exposure parameters to better recognize and take clearer images of pedestrians at night.

Remote Sensing Trends

SPIE publishes Stephen Marshall of University of Strathclyde (UK) talk on trends in remote sensing, including miniaturized multi- and hyperspectral cameras and other devices incorporated into drones and other flying sensors:

Tuesday, October 25, 2016

IHS on Security Market Trends

IHS publishes Top Video Surveillance Trends for 2016 report. Few quotes:

"4K video surveillance has been repeatedly touted as a major trend in video surveillance for the last 18 months and it can sometimes be challenging to see past the marketing hype. Yet make no mistake, the video surveillance market is going to 4K cameras; it’s only a matter of when rather than if. For 2016, IHS is predicting:
• Volumes of 4K cameras shipped in 2016 will remain low, less than 1% of the 66 million network cameras projected to be shipped globally. We are unlikely to see over million 4K network cameras units shipped in a calendar year until 2018.
• More “4K-compliant” cameras will be launched because of the increased use of 4Kp30 and above chipsets, meaning more cameras adhering to 4K standards, such as SMPTE ST 2036-1.
• Like the HD surveillance cameras, early 4K models offered the resolution at lower frame rates. We’ll see more cameras with a higher frame rate offered, and closer ties to other video standards.

eWBM Launches its 2nd Gen Dual Aperture Depth Map Processor: DR1152

PRNewswire: eWBM launches the DR1152 depth map processor, the successor to the DR1151 dual aperture processor.

The DR1152's predecessor, DR1151 (announced last year), accomplished cost reduction in the depth map processor market by implementing Dual Aperture's 4-color 3D image solution technology which extracts RGB and IR signals from a single image sensor. The DR1152 increases its depth level to 121 (nearly 8 times compared to the predecessor). Unlike the predecessor, which works only with 4-color sensors, the DR1152 now supports both 4-color and 3-color image sensors. eWBM believes that the 3-color adaptation opens a new market sector since it removes the barrier of sourcing only 4-color image sensors. This will expedite time-to-market significantly since widely available image sensors can be bought off the self.

The DR1152 supports a depth image resolution of up to XGA (1024 x 768) at 30 fps or 60 fps at VGA resolution. The technologies such as blur channel depth combining, noise reduction, and edge thinning technology are included in the new product. With all these significant improvement, DR1152 contains less gate count compared to its predecessor.

Omnivision Announces Two Purecel Sensors

PRNewswire: OmniVision announces the OV12895, a PureCel Plus-S stacked-die sensor. The 12MP sensor leverages a 1.55um pixel and high-speed architecture aimed to consumer-grade drones, surveillance systems, and 360-degree action cameras.

"We are seeing rapid growth in the markets for consumer drones, surveillance systems, and 360-degree action cameras, in part due to the increasing demand for aerial photography and 4K-resolution panoramic videos in security and virtual reality applications. The 12-megapixel OV12895 aligns well with these consumer product segments because it strikes a balance between solid pixel performance and high resolution, in a widely used 1/2.3-inch optical format," said Kalairaja Chinnaveerappan, senior product marketing manager at OmniVision. "The OV12895 builds on our latest-generation PureCel Plus-S stacked-die architecture and has many desirable features for these applications."

PRNewswire: OmniVision announces latest addition to its PureCel image sensor line, the OV2732. Due to the OV2732's HDR mode, power efficiency and small dimensions, the sensor is suited for compact and ultra low-power surveillance devices.

"The industry is seeing a tremendous increase in demand for IoT-compatible home monitoring systems. Today, resolution at 720p is already considered a mainstream segment, and interests in ultra low-power and good low-light 1080p imaging solutions are now on the rise," said Chris Yiu, senior strategic marketing manager at OmniVision. "Available in a 1/4-inch optical format, the OV2732 sets itself apart by delivering crisp 1080p HD video with advanced features such as frame synchronization and staggered HDR within a cost-effective and power efficient package. It's truly a no-compromise imaging solution for the security market."

Both sensors are currently available for sampling, and are expected to enter volume production in Q1 2017.

ams Acquires Heptagon

Austria-based ams is to acquire 100% of the shares in Heptagon dealing with optical packaging and micro-optics, that, in turn, has acquired MESA imaging an couple of years ago. Heptagon’s headquarters and manufacturing are based in Singapore while its R&D center is in Rueschlikon, Switzerland, and Silicon Valley, USA. The company has over 830 employees including around 120 engineers and 500 manufacturing staff. Heptagon's IP portfolio, primarily in optical packaging, includes more than 250 patent families.

Heptagon’s current 12 month revenue run rate is around USD 90m at negative operating profitability due to current underutilization of production capacity. Heptagon expects substantial revenue growth over the coming years starting mid-year 2017, based on its existing revenue and capacity pipeline and customer commitments. To prepare for this expected growth, Heptagon has already embarked on a major expansion of its Singapore manufacturing capacity with a total capital investment of more than USD 250m in 2016/2017. The expansion is based on a confirmed customer commitment for usage of the additional capacity and is fully funded from existing cash in the business, requiring no funding by ams.

The transaction combines an upfront consideration in cash and shares with a substantial deferred earn-out consideration. The upfront consideration includes USD 64m in cash from available funds, a capital increase of 15% of outstanding shares from authorized capital (excluding subscription rights) and shares from currently held treasury shares for a total value of the upfront consideration of approx. USD 570m. The earn-out consideration will be contingent on future results of Heptagon’s business over fiscal year 2017 with a potential maximum value of USD 285m. Following the upfront share transaction, current Heptagon shareholders which include financial investors, management, and employees are expected to hold around 20% in ams. The transaction is expected to close within the next three months. estimates the total acquisition price at up to $919 million, or approximately €845 million, if Heptagon hits the targets set for its next fiscal year. "Heptagon has raised tens of millions over the years, including from GGV Capital, Innovations Kapital of Sweden, Innovacom of France, Nokia Growth Partners, High Tech Private Equity, Credence Partners, Jolt Capital, AAC Technologies and Heliconia Capital Management."

Alexander Everke, CEO of ams, commented on the transaction, “Combining ams and Heptagon creates the clear #1 in optical sensing technologies and fast-tracks our innovation capabilities. As a result, we expect ams to drive the optical sensing agenda in the years to come and broaden its market reach. Together with our leadership position in our other sensing focus areas Environmental, Imaging, and Audio, this strategic transaction is going to transform ams into the global leader in sensor solutions.

Omnivision Announces Low Power 1/3-inch 1080p30 Sensor for Security Applications

PRNewswire: OmniVision announce OV2735, a 1080p30 image sensor for mainstream security and surveillance cameras. "The 1/3-inch 1080p sensor is the most popular format in the mainstream surveillance camera market. Given its excellent performance and advanced features, the OV2735 is one of the most competitive drop-in camera solutions available for this segment," said Chris Yiu, senior strategic marketing manager at OmniVision. "The OV2735 leverages our proven OmniPixel3-HS™ technology to deliver low-light sensitivity in a more compact and power-efficient package."

The OV2735 is currently available for sampling, and is expected to enter volume production in Q1 2017.

Monday, October 24, 2016

Movidius VPU to Appear in Hikvision Smart Cameras

Movidius’ Myriad 2 Vision Processing Unit (VPU) technology will be powering a new lineup of Hikvision smart cameras. “Advances in artificial intelligence are revolutionizing the way we think about personal and public security” says Movidius CEO, Remi El-Ouazzane “The ability to automatically process video in real-time to detect anomalies will have a large impact on the way cities infrastructure are being used. We’re delighted to partner with Hikvision to deploy smarter camera networks and contribute to creating safer communities, better transit hubs and more efficient business operations.

Thanks to Deep Neural Networks and stereo 3D sensing, Hikvision has been able to achieve up to 99% accuracy in their advanced visual analytics applications. Some of these applications include: car model classification, intruder detection, suspicious baggage alert, and seatbelt detection. The Myriad 2 platform allows these functions to now be processed instantaneously onboard the camera, rather than being sent to the cloud for processing.

There are huge amounts of gains to be made when it comes to neural networks and intelligent camera systems” says Hikvision CEO, Hu Yangzhong. “With the Myriad 2 VPU we’re able to make our analytics offerings much more accurate, flagging more events that require a response, while reducing false alarms. Embedded, native intelligence is a major step towards smart, safe and efficiently run cities. We will build a long term partnership with Movidius and its VPU roadmap.

Update: One of the cool looking Hikvision smart cameras with Movidius technology inside:

Framos Compares ON Semi Automotive Sensors

Framos publishes EMVA 1288 comparison of ON Semi older AR0132 and new BSI AR0136 automotive HDR CMOS sensors, both are 1.2MP and 3.75um pixel: