Glossary of Terms
Commonly used terms in reference to machine vision, digital imaging, CCD cameras, CMOS cameras, and firewire (1394) and Gigabit Ethernet (GigE Vision) interfaces associated with such cameras.

A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

1394: Another name for Firewire (see below).

1394b-Optical: 1394b is the latest version of Firewire and 1394B-optical is a version of 1394B that provides for optical cable as the hardware connection. Prosilica has adopted 1394b-optical for its CV1280F and CV640F products.

1394b-over-fiberoptic: same as 1394b-Optical (see above)

Acquisition: Image acquistion refers to how a computer gets image data from a camera into the computer.

Analog: Analog cameras do not have a digital output. These cameras generally provide a TV-like signal that needs to be digitized in the host computer if it is to be used in machine vision. Although analog cameras are still used widely in machine vision they are quickly being displaced by digital cameras such as Prosilica's 1394 cameras which provide a much higher performance machine vision solution. When comparing analog vs digital cameras, the main differences are image quality, exposure control, speed, and ease of integration.

Area scan: Area scan refers to a camera sensor consiting of a rectangular array of pixels. Area Scan cameras are sometimes called matrix cameras. By way of contrast, Line Scan cameras are those with a sensor comprising a single line of pixels (linescan camera). Currently, all of Prosilica's cameras are areascan devices.

Autoiris (Auto Iris) : Some lenses, particularly those used in outdoor imaging, incorporate a galvanometer-type drive to automatically control the aperture, or iris, of the lens. There are basically two types of auto-iris: DC-type and video type. Prosilica's GC-Series GigE Vision cameras can operate the video-type auto-iris.

Binning: Binning is the technique of combining pixels together on a CCD to create fewer but larger pixels. True binning combines charge in adjacent pixels in a manner that increases the effective sensitivity of the camera. Machine vision cameras do not generally have true binning functions. However, Prosilica's CCD-based cameras have a wide range of binning functions.

Blob Analysis: a machine vision computer algorithm that identifies segmented objects according to geometrical properties such as area, perimeter size, color, etc.

Brightness: In reference to cameras, an offset setting applied equally to all pixels regardless of the pixel value. Similar to the brighness setting on a typical computer monitor or television. See "Offset"

Cameralink: Cameralink is one of the common digital camera hardware interface in the market today. It offers high-data transfer rates, but is seriously limited by cable length and does not have a standard communications protocol. Cameralink is largely being displaced by more modern high-performance digital interfaces such as Gigabit Ethernet (GigE Vision). See the following note: Firewire vs USB2. Compare GigE Vision and Firewire.

CCD: CCD is an abbreviation for charge-coupled device. A CCD sensor is a ligh-sensitive semiconductor device, which converts light particles (photons) to electrical charge (electrons). CCD cameras are one of two dominant types of sensor technolgies used in machine vision. The other sensor technology is called CMOS.

CMOS: Complementary Metal Oxide Semiconductor. CMOS refers to an image sensor technology that is manufactured using the same processes as computer chips. This technology works like a photodiode where the light 'gates' a current that that is representative of the amount of light impinging on each pixel. This differs significantly from CCD technology. There are a number of advantages in using CMOS sensors over CCD including cost, speed, anti-blooming, and programmable response characteristics (ie. multiple slope response). CCD's also have certain advantages.

Compact Vision System (CVS): an industrial computer designed for machine vision applications that is manufactured by National Instruments. NI's official name for their compact vision system is variously "NI CVS-1454", "NI 1455" , "NI 145x". The NI CVS is configured to operate 1394 cameras using Labview IMAQ1394 driver. The National Instruments CVS also has special machine vision features like advanced trigger I/O that match the back panel of Prosilica's CV-series cameras.

Dark Current: Dark current is the accumulation of electrons within a CCD or CMOS image sensor that are generated thermally rather than by light. This is a form of noise that is most problematic in low light applications requiring long expsosure times.

DCAM: DCAM or IIDC is a software interface standard for communicating with cameras over Firewire. It is a standardized set of registers etc. If a camera is DCAM compliant then its control registers and data structures comply with the DCAM spec. Such a camera can be truly plug-and-play in a way that other cameras are not. All of Prosilica's firewire cameras are DCAM-compliant (IIDC 1.30 and IIDC 1.31).

Decibel or dB: A logarithmic unit of measure. When used of digital cameras this unit is usually used for describing signal-to-noise or dynamic range.

Depth of Field: Depth of field refers to the in-focus region of an imaging system. When using a lens, especially in close proximity, objects at and near a certain distance will be in focus whereas other objects in the field of view that are closer or farther away will appear fuzzy, or out of focus. The depth of the region that appears in focus is called the depth of field. Generally speaking, the depth of field will be large if the lens aperture is small (large f-number), and the depth of field will be small with a wide aperture (small f-number).

Digital Imaging: Refers to the capture of a video image in such a way that the resulting image data is in digital format useful for analysis by a computer.

Dynamic Range: The ratio of the maximum signal relative to the minimum measurable signal often measured in decibels or dBs. Dynamic range is sometimes used interchangably with SNR. It can also refer to Optical Dynamic Range.

Extended Dynamic Range: Prosilica's CMOS sensor devices all employ a feature called multislope or extended dynamic range. Multislope provides a non-linear mode that allows the camera to image very bright and very dark detail in the same image frame. This is not possible with CCD-type sensors.

Exposure Time: This is the amount of time that the sensor is exposed to the light. This is the control that is used first (before gain and offset) to adjust the camera. In Labview, the shutter controls are a little confusing: there are ‘manual relative’, ‘manual absolute’, “One-push’ and “auto’ controls. Normally, you should use ‘manual absolute’ where each unit corresponds to 1 us of exposure time. When using the ‘relative’ controls, the units are different – 20us per unit. This control is called "shutter" in Labview and some DCAM controls.

Fast Lens: A lens that admits a lot of light. A lens with a low F-number. A typical fast lens will have a F-number of less than 1.2

Firewire: A standard computer interface and its various versions otherwise called IEEE 1394, IEEE-1394a, or IEEE-1394b. It is an especially fast serial interface that is low cost with plug and play simplicity of integration. It is currently the only interface for digital industrial cameras that is standardized both in hardware and software communications protocols.

Filter Driver: With respect to Gigabit Ethernet cameras, a filter driver, or "filter" is used to reduce the CPU burden when handling large volumes of data coming from Prosilica's GigE Vision cameras. The filter strips out, or "filters", the image data from the Ethernet packets at the lowest level so that the CPU does not have to do this. Using a filter driver can significantly reduce the CPU load associated with image aquisition.

Frame Rate: Frame rate is the measure of camera speed. The unit of this measurement is "frames per second" (fps) and is the number of images a camera can capture in a second of time. Using region of interest readout, some of Prosilica's cameras are capable of thousands of frames per second.

Frame Grabber (or Framegrabber): This is the industry name for the circuit board (usually a PCI card) that is an interface to connect analog cameras, or cameralink cameras, to a computer system. With the wide range of firewire and GigE Vision gigabit Ethernet cameras which do not require such specialized interface cards, frame grabbers are generally no longer required.

Gaging (or Gauging): In reference to machine vision, this is non-contact dimensional examination and measurement of an object using an imaging system or machine vision camera.

Gain: This is the same as the contrast control on your TV. It is a multiplication of the signal. In math terms, it controls the “slope” of the exposure/time curve. The camera should normally be operated at the lowest gain possible, because gain not only multiplies the signal, but also multiplies the noise. Gain comes in very handy when you require a short exposure (say, because the object is moving and you do not want any blur), but do not have adequate lighting. In this situation the gain can be increased so that the image signal is strong.

Genicam: GenICam, or Gen<i>cam, is a camera interface standard from the European Machine Vision Association (EMVA) that offers a software interface that is independent from camera hardware.

Gigabit Ethernet: An industry standard interface, variously called 'gige (gig-ee)', 'GbE', '1000-speed', etc., that is used for high-speed computer networks capable of achieving data transfer rates in excess of 1000 megabits per second. Gigabit Ethernet has been now adapted to high performance CCD cameras for industrial applications. This generalized networking interface is being adapted for use as a standard interface for high-performance machine vision cameras that is called GigE Vision.

GigE Vision: 'GigE Vision' is an interface standard from the Automated Imaging Association (AIA), for high-performance machine vision cameras. GigE (Gigabit Ethernet), on the other hand, is simply the network structure on which GiGE Vision is built. The GigE Vision standard includes both a hardware interface standard (Gigabit Ethernet), communications protocols, and standardized camera control registers. The camera control registers are based on a command structure called GenICam. GenICam seeks to establish a common software interface so that third party software can communicate with cameras from various manufacturers without customization. GenICam is incorporated as part of the GigE Vision standard. GigE Vision is analogous to Firewire's DCAM, or IIDC interface standard and has great value for reducing camera system integration costs and for improving ease of use.

Global Shutter: Generally speaking, when some one says "global shutter", they really mean "snapshot shutter". See "Snapshot Shutter" below. In actuality, a global shutter starts all a camera's pixels imaging at the same time, but during readout mode, some pixels continue to image as others are read out. (see Rolling Shutter, Snapshot shutter). Prosilica's cameras have snapshot shutter meaning that all the pixels start and then stop imaging together. For machine vision applications, snapshot shutter is generally a 'must have'.

Gray Scale: refers to a monochrome image with gradations of grey. An 8-bit camera, for example would represent images in 256 shades of gray. A 12-bit camera would represent images in 4096 shades of grey.

Histogram: A graphical representation of the pixel values in an image. Generally the left edge of the image represents black, or zero, and the right edge represents white, or 256/4096. The histogram curve represents how many pixels of each luminence value.

IIDC: IIDC (DCAM) is a software interface standard for communicating with cameras over Firewire. It is a standardized set of registers etc. If a camera is IIDC compliant then its control registers and data structures comply with the IIDC spec. Such a camera can be truly plug-and-play in a way which other cameras are not. All of Prosilica's firewire cameras are IIDC 1.30 and IIDC 1.31 compliant.

Image Analysis: The software process of generating a set of descriptors or features by which a computer may make a decision about objects in an image.

Integration: generally refers to the task of assembling the components of a machine vision system (camera, lens, lighting, software, etc). Usually used as short form for "System Integration". When used in reference to what the camera does, it is another word for exposure time (see Integration Time).

Integration Time: Also referred to as exposure time. This is the length of time that the image sensor is exposed to light while capturing an image. This is equivalent to the exposure time of film in a photographic camera. The longer the exposure time, the more light will be acquired. Low light conditions require longer exposure times.

Interline Transfer: A CCD architecture where there exists an opaque transfer channel between pixel columns. Such a CCD does not require a mechanical shutter but spatial resolution, dynamic range, and sensitivity are reduced due to the masked column between light sensitive columns. Certain sensors, such as the Sony ICX285 sensor used in Prosilica's EC1380 camera has microlenses on the sensor which mitigate the effect of the masked column.

IR Lens: A lens that is specially designed so that chromatic aberations in the infra-red wavelengths are corrected. An IR-lens should be used in cases where both visible and IR illumination is being received by the camera; otherwise the resulting image would be blurred.

ISO 9000, 9002: Internationally recognized standards that certify a company's manufacturing record keeping. ISO accreditation does not imply any product quality endorsement, but it israther an acknowledgement of the manufacturing and/or engineering record keeping practices of the accredited company.

Jumbo Frames: With respect to Gigabit Ethernet, Jumbo frames refers to the data packet size used for each Ethernet frame. Since each data frame must be handled by the operating system, it make sense to use large data frames to minimize the amount of overhead when receiving data into the host computer. Such large data blocks are called Jumbo frames. To achieve maximum performance from Prosilica's GE-Series gigabit Ethernet cameras, Jumbo frames should be enabled and sized to at least 9000 bytes (although the cameras will also work with smaller frame sizes).

Labview: Labview is a graphical programming language/software application manufactured by National Instruments. It is widely used for testing and control applications and is increasingly being used for machine vision applications. National Instruments also has a number of machine vision-specific software products including Vision Builder for Automated Inspection and the Vision Assistant. Prosilica's GigE and firewire cameras are supported natively by National Instruments vision software.

Linescan (or Linear Array): A linescan, or linear array camera has a single row of pixels and captures an image by scanning an object that moves past the lens. Conceptually similar to a desktop scanner (compare "area scan").

Link Aggregation : Certain Gigabit Ethernet switches have an interesting feature called Link Aggregation whereby the switch dynamically manages the data bandwidth between multiple NIC cards. This is very usefull in some cases. Consider three fast GigE Vision cameras running through a switch to two NIC cards. Without link aggregation, you would need to allocate two of the three cameras to one NIC and the ramaining camera to the other NIC. This obviously does not take advantage of the total bandwidth provided by two NIC cards. However, with Link Aggregation, the switch will optimally divide the data created by the three cameras and equally allocated the data between the two NIC cards thus maximizing the data throughput. Call Prosilica for further details.

Machine Vision: Machine vision is the application of cameras and computers to cause some automated action based on images received by the camera(s) in a manufacturing process. Generally, the term "machine vision" applies specifically to manufacturing applications and has an automated aspect related to the vision sensors. However, it is common to use machine vision equipment and algorithm outside of the manufacturing realm.

Megapixel: Refers to one million pixels - relating to the spatial resolution of a camera. Any camera that is roughly 1000 x 1000 or higher resolution would be called a megapixel camera.

Manual Focus: Refers to a lens which requires a human user to set the focus as opposed to an auto-focus lens which is controlled via a computer or camera.

Manual Iris: Refers to a lens which requires a human user to set the iris as opposed to an auto-iris lens which is controlled via a computer or camera.

Microlens: A type of technology used in some interline transfer CCD's whereby each pixel is covered by a small lens which channels light directly into the sensitive portion of the CCD. Prosilica's CCD products, the EC1350 and EC1380 cameras both have microlenses on the sensors.

Morphology: The mathematics of shape analysis. An algebra who variables are shapes and whose operations transform those shapes.

Motorized Lens: A lens whereby zoom, aperture, and focus (or one or more of these) are operated electronically. Usually, a computer operated controller is used to drive such lenses. The controller often has an RS-232 port through which a camera, or computer, controls the lens.

NI: National Instruments. National Instruments is a large company known especially for its Labview software.

National Instruments: A large company in the scientific and industrial control market that is also a significant player in the machine vision market. See "Labview".

Network Adaptor - another word for the Ethernet interface card or port used found on many computers. With reference to Prosilica's GE-Series gigabit Ethernet cameras, the Network Adaptor refers to the gigabit Ethernet port or PCI card located in the host computer. Also called a NIC, or Network Interface Card.

OCR: stands for Optical Character Recognition and refers to the use of machine vision cameras and computers to read and analyse human-readable alphanumeric characters to recognize them.

OHCI: (Open Host Controller Interface) describes the standards created by software and hardware industry leaders--including Microsoft, Apple, Compaq, Intel, Sun Microsystems, National Semiconductor, and Texas Instruments--to assure that software (operating systems, drivers, applications) works properly with any compliant hardware. In order to operate Prosilica's firewire cameras, the firewire interface in the host computer must be OHCI-compliant.

Offset: This is just the same as the brightness control on your TV. It is a positive DC offset of the image signal. It is used primarily to set the level of “black”. Generally speaking, for the best signal, the black level should be set so that it is near zero (but not below zero) on the histogram. Increasing the brightness beyond this point just lightens the image but without improving the image data. The factory setting for brightness on the EC1380 is generally optimal and should only be changed if it solves some lighting problem. The camera units for brightness are 0 to 255 in linear increments. In Labview, we normal turn this control to “off” or “ignore”. Unless your application really requires this control to be exposed, I would hide it.

Pixel: An abbreviated form of picture element. The individual elements that make up a digitized image array.

QImaging: a digital camera company founded by Prosilica founders Marty Furse, Brian Pontifex and one other person. QImaging focuses on scientific imaging applications, especially microscopy.

Readout: Readout refers to how data is transferred from the CCD or CMOS sensor to the host computer. Readout rate is an important specification for high-resolution digital cameras. Higher readout rates mean that more images can be captured in a given length of time. Prosilica readout rates are faster than most cameras in their class.

Region of Interest: Region of interest readout (ROI) refers to a camera function whereby only a portion of the available pixels are read out from the camera. For example, it is possible to read out a 10 x 17 pixel rectangular area of pixels from our CV1280 camera that has a total resolution of 1280x1024. The result is a much faster frame rate and less data to be processed. This is also referred to as "partial scan" or "area of interest" (AOI). Every Prosilica camera has this function.

Rolling Shutter: Some CMOS sensors operate in "rolling shutter" mode only so that the rows start, and stop, exposing at different times. This type of shutter is not suitable for moving subjects except when using flash lighting because this time difference causes the image to smear. (see Global Shutter, Snapshot Shutter).

Sensitivity: A measure of how sensitive the camera sensor is to light input. Unfortunately there is no standardized method of describing sensitivity for digital CCD or CMOS cameras, so apples-to-apples comparisons are often difficult on the basis of this specification.

Smart Camera: Sometimes called "intelligent camera", or "smart sensor", the term smart camera refers to a camera combined with a low-power computer in a single compact package capable of doing some simple machine vision tasks.

SMB: A simple connector (the gold-colored ones on the back of the CV-Series cameras) that is used for triggering and synchronization. The CV-Series cameras have two SMB connectors (Trigger-in, and Sync-out) that correspond to SMB connectors on on the NI 145x compact vision system.

Snapshot shutter: Sometimes called a global shutter, snapshot shutter refers to an electronic shutter on CCD or CMOS sensors. A snapshot shutter as found on all Prosilica's cameras is a feature of the image sensor that causes all of the pixels on the sensor to begin imaging simultaneously and to stop imaging simultaneously. This feature makes the camera especially suitable for capturing images of moving objects. (see Rolling Shutter, Global Shutter).

Smart Camera: a camera with a built-in computer running image processing software. The output of the camera is usually a text description or an electrical signal instead of image data. Some manufacturers misleadingly use the term to refer to cameras with programmable exposure settings etc rather than cameras running image processing algorithms.

Spatial resolution: A measure of how well the CCD or camera can resolve small objects. Usually used relating not only to the pixel resolution, but also to lens resolution -- ie the resolution of the whole optical system. See also the following article: High Resolution.

System Integrator: A company or person who provides turnkey vision systems using cameras, computers, software, and possibly robotics and other mechanical harware usually aimed at a specific customer application and installation.

Sync: Refers to an external signal generated by a camera than can be used to sychronize the camera with outside events such as flash illumination, or other cameras.

Template: When talking about National Instruments Labview, a collection of Virtual Instruments (VIs) configured for a specific application or function.

Trigger: An input to an industrial digital camera than initiates the image capture sequence. Otherwise, an electrical signal or set of signals used to synchronize a camera, or cameras, to an external event.

Video-type autoiris: There are two major types of auto-iris lenses: DC-type, and video-type. The video-type auto-iris requires a video signal to determine how far to open the iris on the lens. Prosilica's GC-series Gigabit Ethernet cameras support video-type auto-iris. The GC-Series are digital cameras and simulate a video signal on one of the camera outputs which is used to driver the video autoiris lens. The camera software controls also include features to control how the autoiris works. (see Autoiris)

Virtual Instrument (VI): A VI is a set of instructions, or software, that run processes in National Instruments Labview software. (see Template)

Originally posted 2 May 2005, Last edit posted 5 June 2007