Digital video pushes the embedded-technology envelope

For years, video has been an integral part of specialized industrial systems, such as surveillance and factory inspection. However, it has only recently become a viable addition to a range of embedded products. Specifically, the proliferation of image-capture and -playback features in high-volume consumer electronics, such as mobile phones, has produced an abundance of low-cost hardware and software tools to simplify digital-video integration. This widespread availability of digital video in consumer devices has also raised the expectations of embedded-system customers who are no longer content with a user interface that simply displays static information. Users want the same performance from an embedded system as they get from a desktop computer or even a low-cost portable video player. Unfortunately, designers must pay a price to join the digital-video revolution. Designers may have to expand built-in storage, increase processor power, redefine the networking bandwidth, re-evaluate display characteristics, and possibly upgrade the real-time performance of potential products.

Although most consumer-device videos are strictly for entertainment, a multimedia presentation in an embedded device can serve a number of functions. For example, a video display may spice up a mundane product or simplify an overly complex device. Designers can use digital video to differentiate products from the competition and create a unique theme for an entire product line. A video- or graphics-based user interface, along with a network connection, allows remote software upgrades and functional modifications. Built-in training and troubleshooting videos may eliminate or reduce initial installation costs and service calls. With high-bandwidth streaming-video features, customer-support personnel can also interact in real time with customers to solve operational problems or isolate defects.

There are many variations of digital video, and the right choice depends on the system resources, networking capabilities, and operational modes of the embedded device. Obviously, the data volume and bandwidth required to drive a high-definition television are orders of magnitude higher than those necessary to activate a mobile phone’s 128×l60-pixel screen. Most digital-video strategies will impose new hardware and software requirements on the embedded system. An operating system with real-time capability, built-in multimedia features, and device drivers will save on development time. In most cases, you must compress video data for transmission over a communications channel or for local storage. A high-speed networking connection to the Internet or local servers is necessary to stream remote video data in real time.

Data squeeze
One of the earliest design considerations in any embedded-video-system design is to provide the computing resources necessary to run compression and decompression algorithms or codecs. For example, a typical uncompressed television-video stream requires a data rate of more than 20 Mbytes/sec and more than 36 Gbytes to store a half-hour segment. Depending on the algorithm in use and the image content, video compression can reduce bandwidth and storage requirements by a ratio of approximately 30-to-1. Typical compression algorithms for video work by dividing the image into small blocks and then transforming each block into a frequency-domain representation. After transmission or storage, an inverse-cosine transform returns the frequency coefficients to the 8×8-pixel image block. Similar in processing requirements, both the forward- and the inverse-cosine transforms require only a few hundred instruction cycles on a typical DSP. Two principal organizations define and maintain today’s popular image- and video-compression standards. The ITU (International Telecommunications Union) specializes in telecommunication applications and sup-ports the H.26x standards for video telephony. The ISO (International Organization for Standardization) focuses on consumer applications, such as MPEG standards for video.

In addition to codecs, media-centric embedded systems may include DRM (digital-rights-management) software to process some copyrighted material. DRM schemes attempt to enforce copyright or other usage restrictions that the copyright owner defines. For example, DRM may limit where, when, or the number of times that a user can reproduce material. The compressed-data side of the encoder or decoder uses DRM where the data rates are lower. Although details are usually secret, most DRM algorithms are less complex and easier to implement than the codec. The Microsoft Vista operating system contains the PVP (protected-video-path) system that can prevent DRM-restricted content from playing while unsigned software is running. PVP can also encrypt information during transmission to the monitor or the graphics card, which makes it more difficult to make unauthorized recordings.

The display screen is the focal point for any embedded-video system, and designers base most of today’s products on active-matrix LCD (liquid-crystal-display) screens or the more recent OLED (organic-light-emitting-diode) technology. LCD screens are currently the most popular because of their low power requirements, reduced weight, image quality, response time, and reliability. LCDs consist of cells of a light-polarizing liquid sandwiched between two perpendicularly polarized glass panels driven by a matrix of TFTs (thin-film transistors). An electric current changes the polarization characteristics of the liquid and blocks light transmission though that cell. OLED displays are gaining popularity in embedded-system applications because the technology offers potentially brighter, higher contrast images with less power and lower manufacturing costs. These displays rely on organic compounds that a multilayer-printing process deposits in rows and columns onto a flat carrier. Unlike the traditional LCD, OLED displays require no backlight, and they offer a much faster response. One possible drawback of OLED displays is the reduced lifetime of certain color organic materials. Sony recently unveiled a flexible, full-color OLED display that it based on organic-TFT technology (Figure 1). OLEDs typically use a glass substrate, but Sony developed new technology for depositing organic TFT directly onto a plastic substrate, producing a thin, lightweight, and flexible full-color display. The 2.5-in., 0.3-mm-thick prototype display supports 16.8 million colors at a 120×160-pixel resolution.

Smart clicks
Automated-inspection applications, such as part location, package inspection, and assembly verification, typically require significant video processing to analyze real-time images. To simplify this process, National Instruments recently introduced the NI 1722 and NI 1742 smart cameras, which combine an industrial controller, an image sensor, and vision-analysis software to return inspection results instead of images (Figure 2). NI ships the cameras with its Vision Builder software, a menu-driven environment for configuring and deploying machine-vision applications without programming. For more advanced applications, the cameras are compatible with the full LabView library of image-processing and machine-vision algorithms, such as edge detection, pattern matching, and optical character recognition. The NI 1722 features a 400-MHz PowerPC processor, and the NI 1742 includes the 533-MHz version. Both cameras feature a monochrome, 640×480-pixel Sony charge-coupled image sensor plus built-in I/O. This I/O includes two optoisolated digital inputs, two optoisolated digital outputs, one RS-232 serial port, and two GbE (Gigabit Ethernet) ports with support for industrial protocols, including Modbus TCP (Transmission Control Protocol). In addition, the NI 1742 includes quadrature-encoder support and built-in LED-lighting drive, which provides as much as 500-mA constant current and 1A strobed current. The NI 1722 and NI 1742 smart cameras cost $1999 and $2499, respectively.

With support for today’s mobile arsenal requiring as many as 70 formats, content providers are turning to transcoders to translate between the various video configurations. Transcoding is the direct digital-to-digital conversion from one codec format to another to fit the target device. It involves decompressing the original data to a raw intermediate format and then re-encoding it into the target format. Texas Instruments now offers a new digital-media processor for video transcoding in media gateways, multipoint control units, and set-top boxes. The new DSP-based TMS320DM6467 DaVinci processor system on chip provides real-time, multiformat-video transcoding. Integrating an ARM926 core and 600-MHz C64x DSP core along with a video coprocessor, a conversion engine, and targeted video-port interfaces, the system claims a tenfold performance improvement over previous-generation transcoders. The TMS320DM6467 processor costs $35.95 (large volumes).

High-bandwidth machine-vision, security, and video-surveillance applications, such as position location, biometric face recognition, and vehicle-license-plate identification, start with data sampled directly from the content provider’s signal source using analog-capture cards, or frame grabbers. Targeting these applications, Adlink Technology recently announced the PCIe-RTV24, a high-speed video-capture card featuring PCI Express technology and supporting four channels of real-time image acquisition at frame rates as high as 30 frames/sec (Figure 3). It accepts standard composite-color or monochrome-video formats, features programmable resolution, and generates bit maps in all popular digital formats. With RGB24, an RGB format with 24 bits/pixel, for example, the output includes an 8-bit pixel value for each of the red, green, and blue colors. The module also provides a watchdog function and offers four digital-I/O signals that you can use for strobe-light control, trigger acquisition, and alarm signals. The PCIe-RTV24 supports Microsoft Windows or Linux and costs $195.

Remote video
Although video features have many benefits for new-product design, the required development budget plus recurring display and signal-processing resources may be too costly for some deeply embedded devices. These projects can still take advantage of digital video by employing a removable user interface that may be a general-purpose device, such as a PC, PDA, or cell phone. A short-range communications link, such as Bluetooth, 802.11, infrared, or even a hard-wired connection, enables graphical interaction with much less development effort and minimal hardware cost. If the embedded device has a built-in networking connection to the Internet, you can also create a remote, video-based user interface that is accessible from any browser worldwide.

Digital video is now commonplace in high-volume consumer electronics and desktop computers, and embedded-system designers feel pressure from customers to emulate the user experience to remain competitive. Video instruction and real-time user interaction promise to replace more of the printed documentation with each new system generation. The transition to video will probably appear first in highly portable embedded devices using the low-cost technology from cell phones and video players. The growing supply of off-the-shelf hardware and software products gives embedded-system designers the tools to incorporate a sophisticated video interface into every system. Improvements in compression algorithms and the availability of high-volume silicon will continue to lower costs and make complex multimedia features easier to justify with each new product generation.



For more information
· Adlink Technology: www.adlinktech.com
· ISO (International Organization for Standardization): www.iso.org
· ITU (International Telecommunications Union): www.itu.int
· Microsoft: www.microsoft.com
· National Instruments: www.ni.com
· Sony: www.sony.com
· Texas Instruments: www.ti.com



Author Information
You can reach Technical Editor Warren Webb at 1-858-513-3713 and wwebb@edn.com.



AT A GLANCE
· Video features in high-volume gadgets give embedded-system designers a new set of low-cost development tools and silicon signal processors.
· Digital video requires complex compression strategies to live within the limited bandwidth and storage limitations of embedded devices.
· Digital-rights-management software may be necessary to process and display copyrighted digital data.
· Embedded systems with digital video heighten complexity with 32-bit processors, real-time operating systems, and megabytes of memory.



Captions

Figure 1 Sony is producing a lightweight and flexible full-color display by depositing an organic TFT directly onto a plastic substrate.

Figure 2 The NI 1742 smart camera includes an image sensor, a processor, and vision-analysis software to return inspection results instead of images.

Figure 3 The PCIe-RTV24 high-speed video-capture card supports four channels of real-time image acquisition at frame rates reaching 30 frames/sec.

Click here for Illustrations:

Figure 1, Figure 2, Figure 3