Electronic Design
The All-Seeing Eye Of Video Surveillance Cuts The Cord

The All-Seeing Eye Of Video Surveillance Cuts The Cord

Video cameras are eveywhere. Since the terrorist attacks in the U.S. on Sept. 11, 2001, as well as the London subway bombings on July 7, 2005, governments and other organizations have been increasing their surveillance efforts with newer technology and better equipment. These technologies also are playing a role in crime deterrence and unmanned aerial vehicles (UAVs).

Video surveillance systems have languished in the analog world for decades. New technology, though, is yielding some amazing new systems. Like most other video these days, the trend is toward Internet protocol (IP), digital, wireless, often humanless monitoring with intelligent software and all the benefits these improvements offer. Today’s systems incorporate a number of new functions:

• Monochrome to color: While the move away from black and white isn’t really a new trend, it is still going on. In fact, most monitoring today is in color.
• Analog to digital: While more than 80% of all surveillance cameras are older analog devices, most new installations are digital.
• Higher resolution: With digital cameras, resolutions have increased from standard-definition (SD) TV standards to multiple video formats including the 720 and 1080 high-definition (HD) standards.
• IP cameras: Most newer HD cameras are also IP cameras, which digitize the signals, packetize them in TCP/IP, and send the results over Ethernet links.
• New compression methods: Digital cameras produce a super-high data rate for HD video. However, new video compression standards have greatly reduced not only the transmission data rate but also maximized storage capability. Motion JPEG, MPEG-4, and H.264 are the most common compression methods.
• Wireless connectivity: Most installations still use coax as the link from the camera to the monitoring stations. But some new systems use existing Ethernet local-area networks (LANs) or new CAT5/6 LANs installed for the purpose. Even fiber is used in high-end systems. For the sheer convenience and low cost, though, wireless has become the link of choice in many new systems or expansions of existing cabled systems.
• Disk storage. Video archiving has moved from VHS VCRs to hard-disk drives. DVRs and special servers store the compressed video for specific periods of time. Large systems use disk arrays with 2-Tbit drives.
• Video analytics: With HD digital video, computers or FPGAs can be used to analyze the captured scenes for intelligence. Video analytics use artificial intelligence algorithms and machine vision techniques to detect motion and recognize objects and even faces. Most experts agree that the only way to deal with the growing amount of video is to partially or completely replace humans in the analysis of captured video by using the intelligence of video analytics.

Continue to next page

There are two basic types of cameras: analog and digital IP. Both come in a wide range of packages. The bullet option is the most common, but dome and pinhole “spy” cameras are popular too. Indoor and outdoor models are available. Black and white cameras are still available, but most today are color.

The basic imaging sensor is either a charged coupled device (CCD) or CMOS. CCD, which is mature and well developed, is the preferred sensor in very high-resolution cameras where speed is less important. CMOS sensor resolution has almost reached that of CCDs but still lags. It provides very high speed where maximum frame rates are needed.

You can also get an infrared (IR) camera with an IR sensor that can literally see in the dark. IR cameras usually have an array of IR LEDs on the front to illuminate scenes or targets that cannot be seen with the human eye. Their range is usually limited to several hundred feet.

The resolution is a key specification in cameras. The most common is old analog standard NTSC video with 525 total lines interlaced (480 visible) at 30 frames per second (fps). PAL, SECAM, and CIF (Common Intermediate Format) are also available. Resolution is quantified as TV lines (TVL) of resolution or just lines of resolution (LOR). Formats of 380 (minimum), 400, 420, and 480 lines are the base lines. Higher-resolution analog cameras provide 560, 600, or 700 LOR.

The popular CIF standard uses a pixel count of 352 by 288 at 30 fps. NTSC CIF, which can be derived from NTSC, has a 352- by 240-pixel count. Quarter CIF is 176 by 144 pixels. Then, 4CIF offers 704 by 576 pixels, and 16CIF boasts 1408 by 1152 pixels. Some systems use Sony’s D1 video format, which is an SMPTE VTR standard that supports 720 by 480 for NTSC and 720 by 576 for PAL.

Digital IP cameras provide higher resolution. Standard VGA with 640 by 480 pixels is popular. Common pixel resolutions range from about 1 Mpixel up to 12 Mpixels and even more. You can also get cameras that will do 720 or 1080 LOR like standard HD TV, but the price is high. Even higher resolutions are available for special applications.

Tied to resolution is the frame rate. While 30 fps is considered real time, it generates tons of video to store or transmit. Some cameras can provide lower frame rates of 15 or 7.5 fps to save on storage space. Frame rates down to 1 fps are available, but critical action may be missed with such a low rate.

The lens and the physical controls should be considered as well. Lens specs vary and depend on the application, whether it’s close up, long range, or a special angle of view. Some lenses have a motor for remote iris and focus control. Then there are fixed-mount cameras and cameras with motorized pan, tilt, and zoom (PTZ). You can even buy fake cameras with a blinking red LED to make people think they are on TV.

Many cameras include a microphone for audio pickup and motion detection circuitry that turns the camera or recorder on only if motion is detected. For power, most cameras operate from a 12- or 24-V dc source. A few may be battery powered. IP cameras connected to an Ethernet LAN can use Power over Ethernet (PoE).

Continue to next page

As for connectivity, analog cameras use a standard IEEE video signal over 75-O coax, either RG-59 or RG-6. The range is about 300 feet maximum. Analog video can be sent over UTP like CAT5, but it is not common. IP cameras have an RJ-45 connector and a 10/100 Ethernet port (Fig. 1) with a range up to 100 m. Most IP cameras are actually small webservers with their own IP address. You can monitor such cameras from anywhere via the Internet.

The output from the CCD or CMOS imager in an IP camera is already digital. The newest wide dynamic range (WDR) CMOS sensors simply stream the output in a digital format called RAW/Bayer, which uses an RGGB color scheme. The output can be as many as 20 bits/pixel. Data rates soar for high definition at fast frame rates. For example:

20 bits/pixel × (1280 × 720) pixels/ frame × 60 frames/second = 1.10592 Gbits/s

The core of the design is a fast processor with both flash and SDRAM and the video coding and compression. The design also includes a variety of interfaces and the Ethernet, usually 10/100, but some include 1-Gbit/s Ethernet. Most designs use a single system-on-a-chip (SoC), either an FPGA or special DSP, to perform all the functions. These usually include the interfaces. Memory is in separate chips (see “Video Surveillance Compression, Bandwidth, And Storage”).

Wireless Monitoring
While the latest trends in resolution, IP connectivity, and analytics make video surveillance better than ever, wireless technology has contributed more to its ubiquity than anything else. Cable installation is costly, time consuming, and generally off-putting. Also, cables sometimes limit where cameras can be positioned for the best visibility.

A wireless link can solve these problems. With a good wireless modem at each camera, connectivity can be established essentially from anywhere back to a central collection point. Installation is super fast and easy, and the cost is orders of magnitude less than running cables. What makes wireless so viable is simply that the technology is now available to support the high data rates needed over reasonable distances.

Video cameras can use one of three basic wireless topologies: point to point (PTP), multipoint-to-point (MTP), and mesh. In small, simple systems, a direct PTP often is the solution. Some systems can be made to support multiple PTP connections. In MTP, multiple cameras use a single wireless collection point where the data is aggregated, stored, or otherwise processed. When cameras are out of range of a master MTP master node, though, mesh solutions may be used.

Wireless mesh networks have been in development for years, and many excellent systems have been developed. In a mesh system, each node can typically communicate with any other nearby node. Each node acts as a repeater that can send or receive data or relay it to an adjacent node. This relay ability lets you spread cameras and nodes out over a huge range that may generally exceed the range limits of individual radios.

Wireless radios are commonly digital and are designed to transmit Ethernet packets. They are most compatible with the new IP cameras. Yet older analog systems can often be accommodated by using available data converters that digitize and packetize the analog outputs into Ethernet packets that are ideal for wireless transmission.

Continue to next page

The choice of wireless technology depends on many factors such as range, reliability, cost, and required data capacity. Perhaps the most widely used wireless technology is some flavor of Wi-Fi, the popular IEEE 802.11 standard using the 2.4-GHz band or in some cases the 5.8-MHz band. It was designed for Ethernet and quickly and easily matches up with any wired Ethernet LAN.

802.11a/g data rates to 54 Mbits/s are potentially available, with 22 Mbits/s for IP being typical. With 802.11n equipment now available, even greater data rates, exceeding 100 Mbits/s, are possible. The use of multiple-input multiple-output (MIMO) greatly extends range and reliability. Directional antennas also boost the range potential to several miles. Several proprietary 2.4-GHz radios are also available.

900-MHz radios for video are less common but can be used where extra range is needed. Frequency shift keying (FSK) or simple direct sequency spread spectrum (DSSS) modulation makes them very reliable in noisy areas or where interference from many other 2.4-GHz sources may be present. 900-MHz radios are very popular on UAVs because of their robust nature.

Another possibility for faster and longer-haul needs is WiMAX. This broadband wireless technology can handle fast data for many miles reliably. WiMAX systems typically are licensed and use the 2.3-GHz band in the U.S. The 5.8-GHz band is also widely used in the U.S. Even machine-to-machine (M2M) cellular or microwave/millimeter-wave backhaul systems are an option in special cases.

With regard to range, so much depends on multiple factors such as frequency of operation, transmit power, receive sensitivity, types of antennas, their gain and height, and the environment. Hundreds of feet are easily accommodated in most cases, and a range of several miles can be achieved with special equipment.

The required data rate is also a factor to consider. Requirements ultimately boil down to things like video format, frame rate, type of compression, and the upper limits of the wireless technology. Most wireless gear can easily handle compressed video. Even with CIF format at 30 fps using MPEG4-SP, the maximum data rate needed is only 600 kbits/s. And even with HD video and a reasonable frame rate using H.264, rarely is a data rate exceeding a few Mbits/s needed. All this is well within the range of most available technologies.

Reliability is another issue. Today, if the wireless links are properly designed, reliability is nearly as good as a wired system. The goal is to build in extra fade margin to accommodate weather and changing environmental conditions such as moving objects, trees growing or blowing, and new buildings. Finally, security may be an issue. Using encryption and even directional antennas, existing equipment can meet most security needs.

Most new IP systems are wireless just because installation is faster and easier, not to mention cheaper. Existing wired LANs often can be used to carry the video if the system isn’t too great. Wireless modems can easily extend wired systems. And don’t forget the potential to incorporate existing analog systems. Such hybrids are very common.

Continue to next page

Products and Systems
An amazing number of products is available to support the movement to higher-resolution IP video surveillance systems.

Allied Telesis makes Ethernet switches for video systems. The switch is the key component in a larger system. Most larger systems require a separate Ethernet LAN to keep the heavy video traffic from overloading the IT network. A large surveillance network with data coming from each IP camera at up to 4 to 7 Mbytes would probably overwhelm any company LAN.

In an optimized system, each camera sends its data to the switch, which then communicates with the storage servers as well as the client PCs and monitors that have access to the video streams and the control of the system. The Allied Telesis AT-8100S stackable switches are designed for these video surveillance systems (Fig. 2).

FPGA vendor Altera Corp. is also addressing the video surveillance market with a high-definition IP camera reference design based on a single FPGA. This solution features Altera’s low-cost Cyclone III or Cyclone IV FPGAs and intellectual property from Eyelytics and Apical.

The design is used with AltaSens’ 1080p60 A3372E3-4T and Aptina’s 720p60 MT9M033 HD Wide Dynamic Range (WDR) CMOS image sensors. The goal is to reduce board space and power consumption, increase flexibility, and reduce development time compared to architectures using traditional DSPs and application-specific standard products (ASSPs).

Traditional DSPs and ASSPs don’t have the processing power required to accept the large bandwidth of data from 1080p and 720p WDR CMOS sensors. For example, a full HD raster is 2200 by 1125 pixels by more than 16 bits per pixel by 60 frames per second, resulting in a bandwidth that’s greater than 2 Gbits/s. Altera’s FPGAs can deliver the bandwidth. Altera’s HD surveillance IP camera reference design includes:

• Apical’s ISP incorporating best-in-class WDR processing “iridix” together with advanced temporal and spatial noise reduction
• Apical’s “checkerboard demosaic” core for the Altasens A3372E3-4T WDR mode
• “3A” functions, such as auto exposure and auto white balance implemented in software on Altera’s Nios II embedded soft-core processor
• Eyelytics’ H.264 video encoder, capable of 720-line progressive 30-fps encoding or 1080-line progressive 15-fps encoding in main profile
• Altera’s triple-speed Ethernet media access controller (MAC) intellectual property core

By eliminating the need for DSPs or ASSPs and combining all of these functions into one Altera FPGA, designers can take advantage of the cost and power savings with reduced board space. Altera’s single-chip solution reduces power consumption by more than 50% compared to previous designs.

The bundling of a comprehensive list of intellectual property in this reference design gives designers a head start in camera development, shortening development time by as much as one year. All camera designers have to do is customize the FPGA with their own specific features, such as adding their own software for motion detection and pan, tilt, and zoom control.

Continue to next page

Basler also makes some good modern IP cameras (Fig. 3). They are available in fixed-mount and dome versions. Different models are available with different resolutions such as 640 by 480, 1024 by 768, 1280 by 960, and 1600 by 1200 pixels. All of the models use Sony Wfine progressive-scan CCD image sensors.

Frame rates vary depending on the resolution and compression scheme. They range from 30 fps at the lower resolutions to 4 fps at the highest resolutions. The cameras support Motion JPEG, MPEG-4, and H.264 compression standards. The CCD output is digitized in an analog-to-digital converter (ADC) and sent to an FPGA for initial processing. It’s then sent to an internal dual-core 600-MHz DSP with internal memory of 128-Mbyte RAM and 8-Mbyte flash.

The I/O is standard 10/100 Ethernet and RS-232 if needed. Power is by PoE or 7 to 20 V dc. The multi-encoding features allow multi-streaming. Images can also be scaled. An Area of Interest feature lets users define customized regions within the original field of view. Motion detection with alarms can be implemented as well.

While most HD video surveillance is going wireless or over CAT5/6 Ethernet, there is still a place for coax. It seems a waste to ignore that huge installed base of RG-59 or RG-6 coax infrastructure. It’s possible to send HD video over this coax thanks to the Aviia (Advanced Video Interface for Industrial Applications) HD video over coax system from Gennun Corp. Existing analog CCTV users can upgrade to HD by just switching out the cameras and some other equipment. There’s no need to rewire the facility with CAT5/6 UTP or jump to wireless.

The Aviia transmitter (TX) and a receiver (RX) chipset supports 8-, 10-, or 12-bit component digital video. It provides a serial interface for SD 525i or 625i format and HD 720p at 24, 25, 30, 50, and 60 fps. Other formats provided include HD1080i at 50 and 60 fps or 1080p at 24, 25, 30, 50, and 60 fps.

The TX and RX have a reach of 160 m over RG-59 or 230 m over RG-6 in HD format. The SD reach is longer with 330 m over RG-59 and 440 m over RG-6. A dc power over coax (PoC) feature is designed to provide for remote cameras. An auxiliary data channel with a rate to 6 Mbits/s for remote control is also a key feature. And, the system permits up to eight audio channels to be included in the format.

The Aviia chipsets are based on Gennum’s advanced serializer/deserializer technology. The transmitter is designated the GV7600. The GV7601 receiver incorporates cable equalization technology. These devices use a standard sponsored by the HDCCTV Alliance. This technology was developed for broadcast TV.

The video is transmitted uncompressed and without being encapsulated in TCP/IP. The result is a system where a camera can be plugged into a receiving device, and the video can be transmitted with minimal latency and little if any configuration.

Data rates are 270 Mbits/s for SD, 1.485 Gbits/s for HD, and 2.97 Gbits/s for 1080p at 50/60 fps. The chips are available now, along with a reference design. Other applications include digital signage, machine vision, and video conferencing.

Continue to next page

Intersil’s Techwell division has a wide range of chips that support video surveillance equipment design. For example, its TW6864 and TW6868 multi-channel video capture chips are highly integrated single-chip solutions that support multi-channel real-time video and audio capture via PCI Express x1 interface for PC DVR applications.

Each chip boasts four-high quality NTSC/PAL/SECAM video decoders that convert analog composite video signal to digital component YCbCr data. They also use an adaptive 4H comb filter to separate luminance and chrominance to reduce cross noise artifacts. The TW6864 and TW6868 contain a high-performance proprietary DMA controller that fully uses the PCI Express x1 bandwidth as well, enabling it to transfer video and audio data at optimal throughput rate.

These devices support a wide variety of video applications, including decoding and transferring four-channel real-time D1 video or up to 16 channel non-real-time video simultaneously. They support multiple video display formats such as D1, half D1, and CIF.

By implementing Techwell’s advanced proprietary video-switching technology, the chips possess industry-leading video-locking speed in non-real-time switching mode. In addition, coupled with an external Techwell four-channel decoder, the TW6868 can transfer up to eight channels of real-time D1 video and nine channels of audio output.

The Maxim Integrated Products Mobicam3 IP camera reference design kit is based on Maxim’s MG2580 SoC, which incorporates most of the circuitry needed in an IP cam. It can handle CIF, D1, and 720p video and provide H.264 and M-JPEG encoding to 12 Mpixels. It also includes audio codecs for G.722, AMR, AAC, and MP1/2/3.

The chip integrates an ARM9 with Ethernet and USB I/O. Its basic video analytics include motion detection, trip wire, and image tracking. The Mobicam3 kit also specifies a whole slew of other Maxim linear parts to make the complete camera.

Texas Instruments has a complete line of surveillance video chips, reference designs, and software (Fig. 4). The devices are from TI’s DaVinci line of video processors based on the well-known TMS320 DSPs.

The DMVA1 and DMVA2 video processors target the IP camera space. These SoCs are based on the TI TMS320DM643x DaVinci platform and include a 300-MHz ARM9 core. Both devices include video analytics, including people counting, trip zone, motion detection, camera tamper detection, and streaming metadata. They support H.264, MPEG-4, and M-JPEG compression and can handle D1 at 30 fps or 720p at reduced frame rates. Reference designs based on the Aptina 5-Mpixel CMOS imager and Linux software are available.

The DM365 and DM368 also are video processors on a chip designed for IP cameras but without the analytics. Reference designs are available as well. The DM6435 is a digital media processor for HD IP cameras. A video analytics card based on the DM6435 is also available.

The DM8168 SoC consolidates all of the capture, compression, display, and control functions into a single chip. It can handle multiple channels of simultaneous HD, D1, and CIF video encoding and decoding with support for H.264, MPEG-4, and VC1 at HD resolutions. It includes all the needed interfaces such as Gigabit Ethernet, PCI Express, SATA2, DDR2/3, USB 2.0, MMC/SD, and HDMI and DVI. DVRs represent one example application.

Continue to next page

VideoIQ offers an innovative approach in its iCVR cameras (Fig. 5). SD and HD models are available. HD uses the common 1080p format at 30 fps. The company’s unique approach embeds the storage and intelligence inside the camera itself. The HD sensor produces full-color video that is compressed with H.264 and stored on a 500-Gbyte internal disk drive.

An internal Linux-based computer stores the intelligence software, which includes the video analytics and other algorithms for learning its surroundings and applying common situations to the streamed data. The main benefit of this approach is that it is not necessary to use the Ethernet network connection to stream the video to a server. It’s stored internally.

This is fine since most surveillance video isn’t viewed continuously anyway. It’s stored for later access if needed. Each camera is a complete system unto itself. The connection is via Ethernet back to a monitoring system server. Power is typically via PoE but it can also work off legacy 24 V ac or 12 V dc.

The digital HD makes video surveillance so much better today. It’s a major step forward in video monitoring systems. After all, as Ed Strong of Western Digital asks, “What good is the surveillance if you cannot make out the details that can be used as evidence?”

While HD delivers the detail, you still have to store the video for access later if needed. Thanks to compression and some new high-capacity hard drives, the storage becomes realistic. Western Digital has some new hard drives optimized for the surveillance market.

The Western Digital WD AV-GP line of hard drives now includes a 2-Tbyte capacity device (Fig. 6), the industry’s largest available drive to date. WD AV-GP hard drives with WD GreenPower Technology provide cooler and quieter operation and lower power consumption.

Designed to withstand the stringent environments of the surveillance and security markets, the WD AV-GP hard drives reduce power consumption by as much as 40% over standard drives in their respective classes. The WD AV-GP 2-Tbyte hard drive is perfect for applications such as DVRs and surveillance video recording that demand a higher-capacity hard drive and exceptional reliability. These drives are designed for 24/7 operation.

Xilinx also has software and applications for its FPGAs in video processing. This software works with the company’s Virtex and Spartan FPGA families. For instance, its Spartan-6 Industrial Video Processing Kit includes the LX-150T FPGA development board, DVI/HDMI I/O, an Omnividion OV9715 720P image sensor, and reference designs. The software lets designers program the FPGA for most common image processing jobs like edge enhancement, various image corrections, video scaling, noise reduction, and even video analytics for surveillance applications.

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.