As I write these words, the Mobile World Congress show is underway in Barcelona, Spain. At this yearly event, one of the most important cellular communications conferences, mobile handset and tablet manufacturers and their service provider partners reveal their latest and greatest offerings. And alongside them are the silicon and software providers, unveiling next-generation system building blocks which they hope will show up in handsets and tablets at next year's MWC.
One of the biggest announcements coming out of this year's show, at least so far, is Nokia's model 808 PureView phone. I've written many times in recent months about the potential for cameraphones to render standalone cameras obsolete, as well as about the notable embedded vision development potential implied in the burgeoning still and video image capture capabilities of mobile electronics devices. Although the Nokia 808 will probably not sell in large quantities, due both to its fairly high price point (450 Euros, roughly $600 USD) and its archaic Symbian operating system foundation, it's a leading-edge case study of where mainstream handsets will likely be in short order.
The Nokia 808 contains a 41 Mpixel image sensor (no, that's not a typo), notable not only for its high resolution but also for its relatively relaxed 1.4 um pixel pitch, the latter translating into larger-than-otherwise silicon die size and cost but also to better-than-otherwise low-light performance. Equally compelling to me is Nokia's motivation for going with such a robust image capture foundation. The company had unsuccessfully tried for many years to implement robust optical zoom capabilities into its cameraphone designs, and decided this time around to take a different tack.
The largest resolution still images that the Nokia 808 can capture are 38 Mpixels (7,152×5,368 pixels) in 4:3 aspect ratio mode, and 34 Mpixels (7,728×4,354 pixels) in 16:9 aspect ratio mode. Alternatively, the PureView algorithms combine multiple pixels' data together in creating lower-resolution 8 Mpixel, 5 Mpixel or 3 Mpixel photographs. The resultant oversampling not only improves the per-pixel light sensitivity, it also enhances image sharpness. And, as you "zoom" in to objects at the sensor's center, the input-to-output scaling ratio decreases, until it reaches 1:1. Note, though, that there's no traditional "digital zoom" upscaling, which leads to soft and otherwise artifact-filled results. As Nokia's documentation (PDF) states, "We've taken the radical decision not to use any upscaling whatsoever. There isn't even a setting for it."
With the default 5 Mpixel and 16:9 aspect ratio still image capture settings, the effective zoom range is 3x. Higher resolution still images support a narrower zoom range, lower resolution images a wider range, and the same process applies by extension to video. 720p (1280×720 pixel, i.e. 0.9 Mpixel) video capture, for example, supports a 6x lossless zoom range. I'm excited about the Nokia 808 for a number of reasons, beginning with the implications of such a robust sensor (and accompanying image processor) appearing in a volume consumer electronics device. Can you imagine how accurate an optical character recognition algorithm could be, for example, if it leveraged such a high pixel count foundation? And let's also not underestimate the downward price pressure and upward feature set pressure that the 41 Mpixel image sensor and companion processing SoC, presumably not exclusive to Nokia, will put on today's conventional counterparts.
In other news, we've recently added an archive of past Embedded Vision Insights editions to the website. And, as you may have already ascertained given that the previous edition of Embedded Vision Insights came out exactly two weeks ago, we're moving to a twice-monthly newsletter publication pace. I'd like to thank my partners in the Embedded Vision Alliance for their ongoing new-content submissions, without which this distribution step-up wouldn't be possible. And I hope that you're as enthusiastic about the uptick as I am. Regardless of whether you love it or dislike it, I always welcome your comments. Thanks for your interest and involvement in the field of embedded vision, and for your support of the Embedded Vision Alliance.
Editor-In-Chief, Embedded Vision Alliance
An Introduction to Video and Image Processing System Design Using FPGAs
An Introduction to the Market for Embedded Vision in Security and Business Intelligence Applications
Increased Processor Power Will Push Analytics to the Edge
Looking Up: Where Next For Cloud Based Video Surveillance?
E.Coli: Cameraphone-Based Embedded Vision is Your Fluorescence-Detection Eye
Eye Tracking: Samsung's Open-Source Backing
Learn From an Embedded Vision Academic Pioneer: More Information is Here
Augmented-Reality Glasses: Will Google Bring Them to the Masses?
Gender-Triggered Advertising: Embedded Vision-Based Chastising