Modern video surveillance is a far cry from the early days of closed-circuit TV (CCTV) systems with a simple camera and a monitor. Newer technologies are much smarter and can do a lot more. They’re much more flexible, can capture and record many more images, and provide detailed examinations and explanations of what is being observed. Operators enjoy dynamic and interactive control, which older systems simply lacked because they were “dumb.”
Video surveillance is big business. According to the NPD Group’s In-Stat, revenues for video surveillance equipment will approach $15 billion by 2014. By 2015, revenues from cameras (analog and digital or IP), digital and network video recorders, and IP encoders will reach $16.4 billion.
Assembling The Components
Typically, a video surveillance system consists of cameras that provide video inputs, a video decoder, an audio codec, display and motion controllers, video compression/decompression, a host CPU, and interface output circuitry for recording and networking functions. Many of these functions are available either on a single chip or multiple chips, or a printed-circuit board (PCB), including various functions (Fig. 1).
No single video surveillance system topology can satisfy all applications. Much depends on the size of the area being surveyed, the scene (indoor or outdoor) involved, the level of detail and recognition desired, the amount of lighting required, the number of cameras needed, the camera’s ability to pan, zoom, and tilt (PZT), the amount of information being gathered, the storage capacity needed, the level of networking and remote operator control, the video analytical capabilities, and the overall system cost.
One thing is clear. Modern technologies like digital megapixel cameras and auto-tracking PZT are giving designers of video surveillance systems a larger range of options. The choice, however, becomes more challenging for designing system topologies for large-area outdoor surveillance applications.
According to the SMP Group, standard analog cameras are impractical to use in the surveillance of large open areas. These cameras have very small depths of field, and large areas require more cameras and recorders to provide proper detail levels. Digital megapixel cameras, which provide much higher levels of efficiency than older-generation cameras, yield low overall costs and better surveillance details (see the table).
For example, some input cameras feature intelligence with on-chip and/or on-board signal-processing circuitry. You can also get a video decoder, audio codec, and video encoder on one or more chips. The interface circuitry may be housed on the same chip or chips or board as well. The application dictates the choice of components to use.
The trend in video surveillance systems is a move toward IP networked cameras. Such cameras include not only the video sensing circuitry, but also networking and video processing capabilities. They have their own IP addresses and the necessary computing functions for network communications.
While some video surveillance cameras are digitized to record and view images on a computer and a display, only IP cameras digitize the video inside the camera. Analog, IR, and thermal cameras also are used in surveillance systems, but their outputs must be fed to a video input for conversion to a digital stream. Analog cameras cannot connect directly to an IP network. However, an encoder can be used at the camera’s analog output, allowing it to be transmitted over an IP network.
Fixed analog and PZT as well as IP cameras can be used together in a system. For example, an IR fixed analog camera may be used around the perimeter of a warehouse, while a PZT analog camera can overlook the parking lot. Fixed megapixel cameras may be used inside the warehouse, and a number of fixed IP cameras may watch hallways, entrances, and exits. Systems that use both analog and IP cameras are known as hybrid video systems. Generally, megapixel IP cameras achieve megapixel resolution.
Video-surveillance system OEMs work closely with IC manufacturers of various video and audio functions, as well as CMOS and charge-coupled device (CCD) camera sensor makers (mostly CMOS), top players being Sony, Omnivision, Aptina, Sharp, and Samsung. “CMOS IC image sensors are becoming predominant,” says Manny Soltero, Texas Instruments’ end equipment video surveillance manager. Nevertheless, CCD image sensors like those made by Dalsa Semiconductor, which also makes CMOS imagers, are finding their way into high-end industrial and scientific applications.
TI offers a wide range of video surveillance IC hardware and software solutions, including IP cameras, video analytics, video storage and viewing products, remote video security capabilities, and, as Soltero puts it, “unified solutions” including IP camera reference designs like the DM385, DM8127, and DMVA1/2. TI’s TMS320DM368//65/55 IP cameras use the company’s DaVinci processor. TI plans to update its next-generation DM385 and DMVA reference designs for later inclusion in the camera sometime this year.
“Low-light performance is a challenge for video surveillance systems designers,” says Jacob Jose, marketing manager for TI’s video surveillance processors. This is a factor in higher megapixel cameras. “The higher the number of a camera’s pixel count for the same sensor area, the less light exposure is available per pixel.”
The Axis Communications M1143-L/44-L IP cameras feature built-in IR-LED illumination for optimal day and night surveillance (Fig. 2). The high-efficiency IR light sources provide 50 feet of light. According to the company, they promise a new era of IR light-source longevity with highly efficient H.264 video compression.
“Integrated IR light is an attractive feature because there’s no need to configure or refocus the IR light or buy additional accessories,” says Fredrik Nilsson, general manager of Axis Communications. “Our high-efficiency IR-LEDs are sunk behind the lens to dissipate the heat normally generated by traditional IR light sources that cause noise in the picture and shorten light-source lifetimes. The M1143-L/44-L cameras produce up to 50 feet of light for a minimum of seven years, even if used on a 24/7 basis.”
Input video signals are digitized via the H.264 standard, also known as MPEG-4 AVC. It is the most recent video-compression format standard developed for use in high-definition video circuitry for consumer TVs, DVDs, Blu-ray players, portable wireless devices like Apple’s iPod and Sony’s PSP, and video surveillance systems.
This motion-estimation video coding standard processes each video frame, macro-block by macro-block, consisting of 16 by 16 pixels, and has forward and reconstruction paths. The forward path encodes the bits. The reconstruction path generates a reference frame from the encoded bits (Fig. 3).
TI offers a wide range of H.264 codecs for single-stream, dual-stream, and triple-stream video configurations. They offer frame rates of 15, 21, and 30 frames/s for 720, VGA and QVGA, and SXVGA encoding. They also handle MPEG 3/4 encoding.
FPGA IC manufacturers have gotten into IP video surveillance by offering low-cost reference designs based on Altera’s low-power Cyclone IV FPGA IC (Fig. 4). Jointly developed with Eutecus’ massively parallel Multicore Video analytics Engine (MVE) IP, it allows customers to analyze four D1 480-frame/s video channels simultaneously using a single FPGA. D1 normally refers to video resolutions of 704 by 576 pixels, 704 by 480 pixels, and 720 by 576 pixels.
Intersil-Techwell and Maxim Integrated Products also make H.264IC codecs. The TW2960 four-channel video decoder from Intersil-Techwell supports the new 960H technology. The chip’s patent-pending technology offers video definition up to 960 lines, which is a significant upgrade from the usual 760 lines in other video surveillance systems. It’s used in digital video recorders made by Hikvision Technology.
Also, Intersil-Techwell shines in the development of fisheye image correction technology. Unlike software-based solutions that require PC control, this image-processing capability is integrated into analog IP cameras, both CCD and CMOS. Up to four distinct image areas can be selected for the correction of image-distortion caused by fisheye lenses, all under PZT control.
“This selectable video coding (SVC) approach might come along in the future as an amendment to the H.264 standard,” says Joe Spisak, Intersil-Techwell’s director of marketing for video security and surveillance.
Intersil makes audio/video codecs as well, such as the four-channel TW2851 decoder/multiplexer and display processor and the TW2866/67 four-channel video decoder, in addition to audio codecs that also contain a video encoder. Intersil also offers the TW6816 video decoder with a 66-MHz PCI interface, and the TW6869 multi-channel PCI Express video capture IC with a built-in eight-channel video and audio decoder.
Maxim Integrated Products offers H.264 codecs as low-power system-on-a-chip (SoC) devices that are the first in the industry to operate from a USB power source, according to the company. The MG2580 and MG3500 operate from a 240-MHz ARM processor that together with its peripherals shut down to reduce heat generation and minimize PCB space.
Designers must take care in using SoC ICs, though. “Since the trend is toward the use of greater software within video security systems, using an SoC IC in a codec makes it more difficult for a designer to change the software settings,” says Intersil-Techwell’s Joe Spisak.
Z3 Technology’s Zeuss Z3-MVE family of H.264 video encoders uses TI’s DaVinci platform and DM8169 video processor. As a cost-effective solution for 1080- by 6-pixel applications, it allows customers to quickly integrate the DaVinci technology into their applications.
The Interface Factor
Most video signals are interfaced to the monitoring and storage devices in a surveillance system via cables. This is generally the least expensive and most reliable method. However, the deployment of wires can be impractical or very costly in some places, such as parking lots, fence lines, and remote buildings. Here, wireless transmission of video signals is more cost-effective.
Intersil recently teamed up with Sony and Altronix to accelerate the use of hybrid video surveillance, using Intersil’s Security Link Over Coax (SLOC) technology, Sony’s video camera expertise, and Altronix’s eBridge series of Ethernet over coax adapters that incorporate Intersil’s SLOC technology.
“Our adpaters provide extremely cost-efficient upgrades from analog to IP migration onto a network platform by repurposing the legacy coaxial cable infrastructure,” says Alan Friedman, president of Altronix. The Intersil TW3801/3811 SLOC physical-layer (PHY) IC offers a solution for simultaneous transmission of analog color video blanking and sync (CVBS) video and digital IP data over a coaxial cable up to 500 meters (Fig. 5).
The choice of a wired or wireless interface is a major factor, since more than 90% of video surveillance data is sent over legacy coaxial cabling. Obviously, this factor depends on the specific application involved. The end application also dictates the choice of a standard definition camera that offers up to about 3 Mpixels of resolution versus an IP digital camera that can provide up to 20 Mpixels or more. In many cases, both types of interfaces and cameras work together within a video surveillance system.
"Video Compression and Data Flow for Video Surveillance," Zhengting He, Texas Instruments.
"CCD vs. CMOS," Dalsa Semiconductor.
"Building an IP Surveillance Camera System with a Low-Cost FPGA," Altera Corp.