Global Sources
EE Times-Asia
Stay in touch with EE Times Asia
EE Times-Asia > Controls/MCUs

Cellphones get image makeover

Posted: 01 Jul 2006 ?? ?Print Version ?Bookmark and Share

Keywords:Junko Yoshida? EE Times? imaging? CMOS? CMOS sensor?

Camera phones are handy for an unexpected photo op. But consumers have been loath to give up their standalone digital still cameras (DSCs) and rely on camera phones for capturing "important" images. Image sensor companies and camera phone designers must work their way down a long and formidable to-do list if camera phones are to make competitive headway against standalone DSCs.

Technology challenges include enabling handsets to take better pictures in low light and adding features without compromising the phone's ultrathin profiles or increasing its cost.

Indeed, compared with standalone DSCs, "camera phones have 20 times less space to work with, and they need to cost 20 times less," said Jess Lee, VP for the mainstream-products business at OmniVision Technologies Inc. Those realities have some vendors questioning whether turning a camera phone into a high-end DSC is worth an effort.

Studies have found that "90 percent of camera-phone users never print pictures taken with their camera phones," said Philippe Quinio, marketing director of the imaging division at STMicroelectronics (ST).

More specifically, 7 billion images captured by handsets have never been uploaded or printed, said Rutie Adar, product marketing director at TransChip Inc.

Kodak draws a distinction between the two products by calling camera phones "a photo-capable device" and DSCs "a photo-taking device." If current trends continue, "camera phones will become a niche product," said Nancy Carr, VP of marketing for strategic relationships at Kodak's Consumer Digital Group.

Megapixel pressure
Usability concerns continue to dog the sector. Camera phones still lack push-to-send and push-to-print capability, an issue for many consumers. And as megapixel pressures mount for camera phones, picture quality will become a far more serious issue than it is today.

With advances in image-sensor resolution, 2Mpixel or 3Mpixel image sensors are becoming the norm in midrange camera phones. However, a high-resolution sensor alone doesn't guarantee a better photo. Also imperative is a well-thought-out system-level design that can compensate for lens systems that remain mediocre.

"The reduction in pixel size to 2.2m and eventually to sub-2m obviously presents challenges in terms of reduced low-light sensitivity, and an associated increase in noise and decrease in overall image quality," said Tony Henning, editor of Future Image Inc.'s Mobile Imaging Report. "A good sensor can still produce poor images if the lens is cheap or the raw data is mishandled."

At the heart of the issue lies the system-level design that manages the imaging flow. "Pixel technology, image-sensor processor algorithms that can compensate for bad optics and optimized specs for optics are the three fundamentals you need to own," said ST's Quinio. "Missing one of the three could be a problem."

Last year, the focus was on new pixel architectures to combat the decreases in light sensitivity. This year, the hot topic is autofocus.

Autofocus: Hot topic
Camera phones are typically equipped with small, fixed plastic lenses. "Without autofocus, anything closer than 60cm gets blurred, while anything farther than 20m also blurs," said TransChip's Adar.

At the 3GSM World Congress last February, TransChip demonstrated an autofocus variant for camera phones. But camera phones equipped with autofocus remain a rarity. "Autofocus based on a very simple voice-coil system is too big and bulky for a thin camera phone," said OmniVision's Lee, and "its cost adder is about $4." The technology's ability to clear the stringent drop testas high as 1.5mrequired for camera phones is another concern. The drop test for DSCs isn't as rigorous.

Many image-sensor designers are turning their attention to the disparities between advanced image sensors and the basic plastic optics used in today's camera phones, seeking solutions that can bridge the performance gap.

Guy Michrowski, VP of marketing and sales at Dblur Technologies Ltd, cited two reasons for the industry's passion for new optics solutions. First, he said, "sensors are making gigantic steps forward, while the optics have remained the same." Second, with the recent reductions in pixel size, "image sensors are starting to suffer."

Software lens
The industry must "break away from the physical limitations of optical design," Michrowski said.

Dblur plans to replace the physical lens with a "software lens system" that combines a customized, Dblur-designed physical lens with a proprietary image-processing IP core. The first target for the technique will be autofocus capability, but the company claims the basic technology can likewise be applied to yield superior photo quality, simplified optical zoom or substantially reduced product height for the camera phone.

OmniVision, which acquired CDM Optics last year, promises to deliver a high-volume autofocus feature called OmniFocus for camera phones later this year. By applying CDM's patented optical-encoding technology, called Wavefront coding, OmniVision claims that OmniFocus can decode an optically encoded out-of-focus image to turn it into a sharp image in an increased depth of field.

Liquid lenses such as those from Varioptic are now reaching the market; microlenses from Microalign show great promise; and prism lenses from Olympusalong with piezo-driven actuators from Johnson Electric, 1 Ltd and New Scale Technologiespromise small, light, rugged, power-efficient motors to drive autofocus and optical zoom, Henning said. All of these advances will produce "improvements in camera phones" and will become "major differentiators for the manufacturers who deploy them first," he said.

Meanwhile, pixel design continues to be tweaked to improve sensitivity. Cypress Semiconductor Corp. rolled a three-transistor-per-pixel architecture that provides low-light sensitivity without increasing the pixel transistor count or sacrificing image quality. And Kodak's Pixelux technology combines pinned photodiodes, four-transistor pixels and a shared-pixel architecture to achieve high sensitivity under low-light conditions.

Last year, Micron Technology Inc. demonstrated the first 1.7pixel image sensor. The company believes shrinking pixel size enables camera phones to offer "DSC-like photo quality, camcorder functions and high-definition capabilities," said Bob Gove, VP of imaging at Micron. The increasing availability of Wi-Fi and WiMAX and the rise of removable media in mobile handsets will make it much easier to migrate genuine DSC technologies to mobile phones, Gove said.

Although the details are still under wraps, Advasense, an image-sensor startup funded by Blue Run Ventures is scheduled to start sampling an image sensor with reportedly 10 times better sensitivity performance than existing solutions by the end of this year.

Frame-level memory
Another design factor that separates the image quality of DSCs from that of camera phones is the availability of image memory. In a DSC, a raw image is first grabbed and stored to memory and is then postprocessed to JPEG, ST's Quinio said. In camera phones with no frame memory, raw images are grabbed and processed on the fly, typically using a small line buffer.

Micron Technology believes it's just a matter of time before frame-level memory is integrated into system-on-chip sensors. "Multiple image frames are essential for enabling video functions in camera phones," said Gove.

Meanwhile, the debate over partitioning of the image sensor and the image signal-processing block remains unresolved. Qualcomm Inc. has advocated the separation of image-processing capabilities from image sensors so that baseband processors can provide that functionality. Others have shown image pipeline algorithms packed into an application processor used in a high-end phone.

TransChip's Adar, however, argued that tight coupling between ISP and the image sensor module enables a faster time-to-market for handset vendors. For baseband processor companies whose core expertise is not imaging, developing advanced image-processing algorithms has proved increasingly difficult, she said. Moreover, "despite their investment, baseband processor companies are finding that they can't charge extra" for image processing.

Other image sensor companies, while agreeing with Adar's observation in principle, said that they're obliged to make different versions of image sensors to address camera phones at different price points.

- Junko Yoshida
EE Times

Article Comments - Cellphones get image makeover
*? You can enter [0] more charecters.
*Verify code:


Visit Asia Webinars to learn about the latest in technology and get practical design tips.

Back to Top