Video Transmission for Machine Vision Cameras

During recent years most area scan cameras in machine vision applications conform to one of the common video interface standards. In Europe these are CCIR and PAL and in the United States RS-170 and NTSC. CCIR and RS-170 are monochrome standards while PAL and NTSC are colour.

There are a multitude of standards for transmission of video data. However they can be split into two distinct groups – Analogue and Digital. With a machine vision system the image data is ultimately converted to a digital representation. The fundamental difference between the two transmission techniques is the point at which the digital conversion takes place. With the analogue standards the conversion is done in the frame grabber while with digital transmission the data conversion is performed within the camera itself. We will investigate the analogue standards first.

Analogue Video Transmission

By far the most common flavour of analogue video standards are the American RS-170 and European CCIR standards.

The table below lists the key parameters of each standard:

As you can see the two standards are very similar, the difference in image frequency being attributable to the power supply frequency.

The structure of the transmitted signal (CCIR) is shown in the diagram above. As mentioned earlier all the odd lines of pixels making up the image are read out first followed by all the even lines. Each group of lines is known as a field.

The beginning of each field is marked by a vertical synchronisation pulse (VSYNC) equivalent in time to about 25 lines. Each field is composed of a sequence of lines, each line being separated by the horizontal synchronisation pulse (HSYNC), sometimes known as the line-sync. The position of the VSYNC pulse is used by the receiving device to determine whether the field is odd or even. This form of video signal is also referred to as composite video because the synchronisation signals are embedded in the image waveform. White is defined as 100% full scale while black is 7%.

Digital Video Transmission

There are two main methods of digital video transmission – Firewire and Camera Link.

Firewire is the popular name for the IEEE 1934 bus system and is a serial bus interface standard, offering high-speed communications and isochronous real-time data services. FireWire replaced Parallel SCSI in many applications due to lower implementation costs and a simplified, more adaptable cabling system.

FireWire uses 64-bit fixed addressing, based on the IEEE 1212 standard.
Each packet of information sent by a device over FireWire consists of three parts as follows:–

• A 10-bit bus ID that is used to determine which FireWire bus the data came from
• A 6-bit physical ID that identifies which device on the bus sent the data
• A 48-bit storage area that is capable of addressing 256 terabytes of information for each node

The following table shows the properties of FireWire according to the IEEE 1394:*

Property	Description
Communications Scheme	Serial
Bus Topology	Tree
Number of Devices	63
Transmission Modes	Asynchronous for control signals at around 20% of bus signals and Synchronous for video/audio at around 80% of bus cycles
Configuration	Self-configuring and hot-plugging
Cable	Six wire
Transmission Rate	100, 200, 400 and 800Mb
Data Security	CRC check in each data packet
Bus Cycle	Typically 125µs

*Interesting points to note are the maximum number of devices, along with the hot-plugging configuration means that up to 63 devices can be connected to a single socket at any time and the bus will automatically reconfigure to support the new setup. The suitability of FireWire for image processing applications is clear with the synchronous transmission mode utilising 80% of bus cycles. This means for example on a 400Mbps transfer rate a delivery of 40million 8-bit pixels per second, equal to 100 images of a typical VGA sized image or in excess of 30fps @ 1024 x 1024. Meanwhile the asynchronous mode allows for transmission of data such as results from line-scan or spectroscopy cameras.

Camera Link is a high-speed serial data interface standard based on National Semiconductor's Channel Link interface, shown in the left hand diagram. The transmitter converts the 28-bit data into four Low Voltage Differential Signalling (LVDS) data streams. With every cycle of the transmit clock, the data is sampled and transmitted to the receiver which converts the LVDS data streams back into 28-bit data format.

Camera Link uses the Channel Link interface to define its standard including Basic, Medium and Full configurations as shown in the diagram below.

A single 26-pin Camera Link cable contains up to 24-bits of video data, a clock line as well as enable and control signals. Camera Link data rates can now reach up to 2.3Gbps and as the transmission speeds are so great, cameras using Camera Link need little on-board memory to store captured images. As an example of typical operation a 400Mbps Camera Link can transmit around 30 frames per second at 1280 x 1024 pixels.

The purported benefits of Camera Link are:-

• Easy Product Interchange
  Every Camera Link product will use the same cable connection. Cameras and frame grabbers can easily be interchanged using the same cable.
• Simple Interface
  Only two connections will be required to interface a camera and frame grabber: Power and Camera Link.
• Cheaper Cable Prices
  Because Camera Link is an industry-wide standard, consumers will be able to take advantage of lower cable prices.
• Smaller Cables
  The technology used in Camera Link reduces the number of wires required to transmit data, allowing for smaller cables. Smaller cables are more robust and less prone to breakage.

GigE

Until recently FireWire and Channel Link have dominated the video transmission market. Now GigE excels these previous standards and currently stands as the most technologically advanced interface for high performance digital cameras. GigE is the newest generation of Ethernet technologies. Standard Ethernet began at 10Mbps which improved to 100Mbps Fast Ethernet standard which has now been surpassed by the GigE standard at 1000Mbps or 1Gbps as the name suggests. Until this point the previous standards were just too slow to allow uncompressed video data to be streamed in real-time.

While the GigE standard defines a network structure, the GigE Vision standard actual defines the interface standard for these networks. The ongoing development of this interface standard is overseen by the Automated Imaging Association (AIA) currently offering;

• Real-time transmission of uncompressed video data at segments over 100 metres.
• Allows for any combination of single multiple instances of cameras and computers.
• Low cost standard cables, connectors and easily integrated hardware.
• Highly scalable future development following the rise of Ethernet bandwidth.

GigE Vision encompasses the GigE standard along with communication protocols and standardised methods of communicating with and controlling a camera. The standard is comprised of four main parts; firstly the GigE Vision Control Protocol (GVCP) defines how to control and configure compliant cameras and send image data to the host. Secondly the GigE Vision Stream Protocol (GVSP) defines data types and describes how images are transmitted. Thirdly a Device Discovery Mechanism (DD) defines how cameras and other compliant devices obtain IP addresses. Lastly cameras and devices will be supplied with an XML file based on the GenICam standard, providing the equivalent of a computer readable datasheet to detail access to its controls and image streams.

Comparison of Digital Video Transmission Techniques