Monday, September 01, 2014

Why ISO Changed Resolution Chart

Image Engineering publishes explanations on why ISO changed the resolution chart in Feb. 2014.

"The problems users had with the old chart:
  1. It is defined as a high contrast test chart with a white background. This sometimes causes clipping in the highlights or/and the shadows.
  2. Visual analysis is often impossible
  3. The high contrast edge is subject to sharpening which leads to misleading results when looking at an SFR derived from the edge"

The new chart is sine Siemens star image considered to be better suited for the resolution measurements, although system sharpening can still present a measurement challenge.

Sunday, August 31, 2014

FLIR One Camera Teardown

Mike Electric publishes a video teardown of FLIR One camera and its thermal sensor Lepton core (part 1, part 2):

Saturday, August 30, 2014

Google Glass Power Consumption Dominated by Camera

Rice University group presented a paper on Google Glass power consumption breakup at Asia-Pacific Workshop on Systems 2014 (APSYS) at Beijing, China:

"Draining our Glass: An Energy and Heat Characterization of Google Glass"
Robert LiKamWa, Zhen Wang, Aaron Carroll, Felix Xiaozhu Lin, and Lin Zhong

The power is by far dominated by camera and image processing:

The same is seen in the temperature profiles:

Microsoft Applies for Energy-Optimized Image Sensor Patent

Microsoft-sponsored research on low power image sensor is now published as a patent application US20140232932 "Energy-proportional image sensor" by Robert LiKamWa, Nissanka Arachchige Bodhi Priyantha, Matthai Philipose, Lin Zhong, and Paramvir Bahl. Basically, it says that all the circuits need be to in power down mode during the idle time, and the resolution needs to be lowered when not necessary:

Compressed Sensing Methods Aimed to Power Reduction

Sony and Stanford University files a patent application US20140231620 "Image sensor and imaging method with single shot compressed sensing" by Yusuke Oike and Abbas El Gamal aimed to save the image sensor power: "Low power consumption is a primary concern in many CMOS image sensor applications. As the resolution of these sensors has increased while maintaining or increasing their frame rate, the analog to digital conversion (A/D) associated with the sensors has become a dominant component of power consumption. Typical image compression techniques reduce the readout rate (and hence the I/O power consumption), but cannot reduce the power consumption associated with the A/D conversions."

So, the application proposes to modify the image sensor readout architecture so that compressed sensing techniques can be applied to save power. The idea is to divide the pixel array into small blocks 12 and add a MUX at the ADC input combined with an ADC activation generator "generating a random activation code that determines which of the plurality of A/D converters is activated; and obtaining a digital read-out of the analog signals from the activated A/D converters."

The readout timing is modified, so that for each pixel block 12, first the reset level is digitized and then the signal level goes through oversampled ADC, such as 1st order sigma-delta modulator:

Then the MUX select control PMX and ADC activation code ACT work in concert to distribute the signal across a number of ADCs, whereas the reset and signal level from the each pixel are processed by the same ADC:

"According to compressed sensing a compressible signal can be recovered from a small number of random measurements by sparsity promoting non-linear recovery algorithms. Therein, the number of A/D conversions represented by M can be sufficiently less than the number of pixels in a unit of pixel block represented by N, to recover the original image. When M number of A/D converters are simultaneously dedicated to N pixels, the image sensor outputs M digital codes for the N pixels so that the compression ratio M/N can be sufficiently less than 1.

The power consumption and bandwidth of both the A/D conversion and input-output transmission can be suppressed by a factor of M/N in comparison to a normal operation wherein N, A/D conversions are required to obtain an image of N pixels. For example, 16 A/D converters dedicated for a unit of pixel block of 256 pixels achieves a compression ratio of 1/16. Note that the compression ratio can easily be varied by changing the number of A/D converters dedicated for a unit of pixel block at the multiplexers. Further, this feature also enhances the frame rate for a given power consumption level, as frames are outputted at the rate of N/M

Friday, August 29, 2014

Toshiba Image Sensor Strategy

Business Wire: In quite an unusual PR, Toshiba has published its image sensor business strategy and priorities:

Toshiba (TOKYO:6502) will focus on development and sales expansion of automotive and medical imaging sensor, in addition to sales expansion with Chinese smartphone manufacturers based on the BSI (backside illumination image sensor) technology and technologies that realize high sensitivity and speed. The current image sensor market is mainly for smart phones and digital cameras and has a scale of approximately 800 billion yen [1]. Future growth is expected on such factors as expansion to automotive applications and bringing higher performance levels to existing applications.

Smartphone sensors

The volume zones for image sensor used in the main cameras mounted on current smartphones are 8- to 13-million pixels. Toshiba supplies products mounting high-sensitivity image processing technology such as HDR[2] and Bright Mode[3]. We have already developed a 20-million pixels sensor for high-end market. We are meeting customer needs by developing more cost-competitive products.

Automotive sensors

Currently Toshiba sells sensor for view camera. We will use our original HDR to develop automotive image sensors, as demand for such products is expected to expand in the future. We will propose a system with TMPV75 Series, our automotive image recognition LSI, so as to expand sales to both domestic and overseas customers.

Medical sensors

We have formed an alliance with and invested in CapsoVision, Inc. (CV), maker of a capsule endoscope, as one aspect of promoting Toshiba’s healthcare business. We are now developing image sensor to mount in CV capsule endoscopes. We are also considering entering the market of sensors for catheters or endoscopes.

[1]: Toshiba survey.

[2]: HDR (High Dynamic Range) is a function that captures high contrast scenes naturally by widening the range of light and dark tones. Toshiba has developed HDR for video as well as still images in both high light and dark contrast settings. Our HDR synthesizing realizes smooth and vivid video recording by suppressing blurring and false colors.

[3]: Bright Mode is a technology to realize bright high speed movie recording. It can record movies that are up to 4 times brighter than those shot in normal mode. It can also record at frame rates as high as 240 frames a second in full HD equivalent by adopting an interlaced format. Toshiba provides high quality video technologies required for high speed video and for such features as smooth slow-motion playback, etc.

The new PR joins a series of similar ones on LSI strategy, discrete semiconductors strategy, and storage products strategy.

Omnivision Reports Quarterly Results, Keeps Quiet on Buyout Offer

PRNewswire: During the last fiscal quarter ended on July 31, 2014, Omnivision shipped 249 million sensors. The non-GAAP gross margin was 22% and net income was $54M, bringing the company cash and short-term investments balance to $524M.

Revenues for the quarter were $406.5M, as compared to $331M in the previous quarter, and $373.7M a year ago. The company expects revenues in the next quarter will be in the range of $360M to $390M.

SeekingAlpha publishes the company's earnings call transcript. Few quotes:

Shaw Hong, CEO:

"The Company's Board of Directors is reviewing and evaluating HCM's proposal. No decision has been made with respect to the proposed transaction. And there are no assurance that this or any other transaction will be consummated. We are focused on our operating our [bench] (ph). We do not intend to provide update with respect to this or any other transaction except as required by applicable law. We think you in advance for your understanding."

"We are also developing a full complement of ASIC with advanced image procession functions by combining our sensors with image processing and other peripheral functions. We can offer our customers with complete imaging system solution, such as in our machine vision products. In our core smartphones, tablet and PC market cameras are enabling new application such as capturing 3D depth information using computational photography to improve image quality and developing new human interface applications. We have partnered with industry leaders to enable 3D depth sensing for both smartphones and tablets."

"We are extremely pleased with our success in China and we expect this market will continue to be a major driver of our future goals."

"We expect India to rapidly grow to become a significant geography for OmniVision over the next few years."

Ray Cisneros, SVP of Worldwide Sales and Sales Operations:

In our first fiscal quarter, we shipped 249 million units, as compared to 201 million units in our prior quarter. The increase in volume was predominately driven by increases in 5-megapixel and HD sensors. The average selling price in our first quarter was $1.63 as compared to $1.64 in our prior quarter.

"Unit sales of 3-megapixel to 5-megapixel category represented approximately 40% of total shipments as compared to 34% in the prior quarter for the same category. The significant increase in unit shipments in this category was predominately driven by larger 5-megapixel sensor shipments to the mainstream smartphone segment in Asia."

Dongbu Foundry to be Sold in September

EETimes: Hindustan Semiconductor Manufacturing Co. (HSMC), Semiconductor Manufacturing International Corp. (SMIC) and three private equity groups Hahn & Company, Askveritas and Bain Capital have expressed interest in acquiring image sensor foundry Dongbu HiTek. Dongbu and its main creditor, Korea Development Bank, agreed to select the preferred bidder in September 2014. Some of the past and current Dongbu customers are Pixelplus, SiliconFile, SETi, Clairpixel, Rayence, Foveon.

Thursday, August 28, 2014

iPhone 6 Camera Expectations

Chipworks analyst Ray Fontaine publishes an analysis "iPhone’s Cameras – What’s Coming in the iPhone 6?" based on the history of iPhone cameras:

Himax To Manufacture 1.12um Pixel Sensor at TowerJazz-Panasonic Fab

Business Wire: TowerJazz Panasonic Semiconductor Co. announce its first third party CIS customer, Himax Imaging, that is developing the next generation high end cameras for smart phone applications using TPSCo’s 1.12um pixel. The pixel incorporates cutting edge light loss reduction technologies using a proprietary 65nm process, allowing 1.12um pixel with a QE of 74%. Through proprietary design and processing the dark current at 60C is measured at 6.5 electrons per second, and the readout noise is 0.9 electrons.

"Himax Imaging has been focused on delivering high value image sensors to mobile, consumer, surveillance and automotive markets with our ultra low power and low noise architectures. After extensive evaluation of TPSCo’s 1.12um state of the art pixel, we believe that our architecture combined with TPSCo’s advanced 65nm CIS process, which provides extremely low dark noise, high quantum efficiency, and excellent angular response, will enable DSC quality performance for our next generation of high resolution image sensors," said Amit Mittra, VP and CTO of Himax Imaging.

"We are very pleased that Himax Imaging is putting their trust in TowerJazz and is now working with us through TPSCo to enable technological advances for its next generation products," said Avi Strum, VP and GM, CMOS Image Sensor BU, TowerJazz.

"By utilizing the advanced features of our 1.12um-pixel CIS 65nm process, Himax Imaging will offer evermore innovative technology to meet their customers’ -- some of the world’s leading device manufacturers -- growing demands for both high quality, yet cost-effective, cameras for high end smart phone applications. We are excited to work with Himax Imaging and look forward to more product engagements in the future," said Guy Eristoff, CEO, TowerJazz Panasonic Semiconductor Co.

TPSCo’s 1.12um-pixel CIS platform includes: advanced 65nm specialty process for CIS, color filters and micro-lenses, stacked light pipe technology for enhanced optical performance, optical/device simulations design optimization, operating voltage of 1.2/1.8/2.8V, 4LM metallization (3LM-Cu and 1LM-Al) and 32nm pitch M1. For mobile and DSC applications, the features include: high resolution sensors, demanding frame rate, high sensitivity and dynamic range, low dark current, and high angular response performance.

Reuters adds that, according to its sources, "the deal would bring in $20 million a year in revenue for the [TowerJazz-Panasonic] joint venture"