Embedded Vision Insights: November 15, 2011 Edition

Dear Colleague,

Welcome to the third edition of Embedded Vision Insights, the newsletter of the Embedded Vision Alliance.

This past few weeks have been particularly newsworthy for camera-inclusive smartphones and tablets. Consider, for example, handsets such as the HTC MyTouch Slide 4G and its plethora of "power user" snapshot settings, the 1080p video capture capabilities of the Apple iPhone 4S, the stitch-free panorama mode supported by the Samsung Galaxy Nexus and the high quality Carl Zeiss optics built into the Nokia Lumia 800. Key to new capabilities such as these are the systems' microprocessors; now-sampling CPUs built from Qualcomm's latest Krait and ARM's latest Cortex-A15 microarchitectures, for example, along with Nvidia's in-production quad-core (or more accurately, penta-core) Tegra 3 and Apple's dual-core A5.

To be clear, these systems (and the SoCs they're derived from) are useful for a diversity of embedded vision functions, not just for picture-snapping and videography purposes. Take a look, for example, at the Kinect-reminiscent gesture interfaces supported by Kinectimals for Windows Phone 7, included in latest-generation Pantech handsets, documented in both filed and granted patents from Apple, and suggested by recent Qualcomm acquisitions. Ponder the facial recognition-based unlock capabilities built into Google's "Ice Cream Sandwich" Android v4 and Nokia's Symbian O/S. And appraise the fresh perspectives represented by embryonic applications such as television program identification, augmented reality, and traffic flow optimization.

Cellular handsets and tablet computers are compelling platform for implementing embedded vision, by virtue of the prevelence of both front- and rear-mounted image sensors of sufficient resolution, the substantial available memory and processing resources, the systems' application-enabling portability, and (perhaps most importantly) the often-subsidized prices at which they're sold and their consequent large installed user base. How do you hope to harness mobile electronics' potential in actualizing your embedded vision, and what barriers exist to transforming your aspirations into reality? Drop me an email with your thoughts, and enjoy this issue of Embedded Vision Insights.

Brian Dipert
Editor-In-Chief, Embedded Vision Alliance

FEATURED VIDEOS

A Conversation with Bruce Flinchbaugh
Jeff Bier, Founder of the Embedded Vision Alliance, interviews Bruce Flinchbaugh, Texas Instruments Fellow and Manager of Vision R&D. Bruce, a pioneer in embedded vision whose career in vision technology spans 40 years, talks about the origins of Texas Instruments now-established businesses in automotive vision and video surveillance applications. Jeff and Bruce also discuss the heterogeneous processor architectures used by TI for embedded vision applications, and how TI’s customers harness these architectures to realize their applications.

A Demonstration of an Optical Flow Algorithm on an FPGA
This BDTI project evaluated high-level synthesis tools that use C code (or other high-level languages) to generate FPGA designs. As part of the tool evaluation, BDTI implemented an optical flow algorithm operating on high-definition video using a sub-$30 FPGA.

A Conversation with Goksel Dedeoglu
Jeff Bier interviews Goksel Dedeoglu, Texas Instruments Member of the Technical Staff in Vision R&D. Goksel, who began his career in vision technology as a university researcher (including earning a Ph.D. from Carnegie Mellon University), explains some of the key trade-offs faced by TI in creating its “VLIB” vision software component library. Jeff and Goksel also explore challenges and best practices for embedded vision system development.

More Videos

FEATURED ARTICLES

How Does Camera Performance Affect Analytics?
Camera designers have decades of experience in creating ISP (image signal processor) pipelines which produce attractive and/or visually accurate images. But which elements of an ISP are most important to get right for good embedded vision analytics implementations, and how do they impact the performance of the algorithms which run on them? This article surveys the main components of an ISP, highlights those components whose performance is particularly important for embedded vision applications, and discusses what their effect is likely to be. More

Automotive Driver Assistance Systems: Using the Processing Power of FPGAs
In the last five years, the automotive industry has made remarkable advances in driver assistance (DA) systems that truly enrich the driving experience and provide drivers with invaluable information about the road around them. This article looks at how FPGAs can be leveraged to quickly bring new driver assistance innovations to market. More

Challenges to Embedding Computer Vision
Computers and robots in science fiction are endowed with vision and computing capabilities that often exceed those of mere humans. But researchers understand the truth: computer vision is very difficult, even when implemented on high-performance computer systems. This article extrapolates computer vision to the world of embedded systems, where even more challenging (as well as unique) design constraints must be surmounted in order to create successful products. More

More Articles

FEATURED FORUM DISCUSSIONS

Useful Resource for Info on Face Recognition

Anyone using OpenCV on Android?

iPhone/iPad OpenCV Links

Experiment with OpenCV on Windows – No Programming Required

Mass Effect 3 Support For Kinect

More Forum Discussions

If you're building AI or vision-enabled products, you've come to the right place.

Embedded Vision Insights: November 15, 2011 Edition

Pages

Topics

Contact

Address

Phone