Camera & Accessories Search

Tuesday, February 5, 2008

Glossary of Image Analysis Terms - N to Z

Glossary of Image Analysis Terms - N to Z


N

Neural Networks – A computing paradigm which processes information based on biological nervous systems. No programming is involved as in artificial intelligence. Rather, decisions are made based on weighted features analyzed by interconnected nodes of simple processing elements using analog computer-like techniques.

Noise – Irrelevant or meaningless data resulting from various causes unrelated to the source. Random, undesired video signals.

Noninterlaced – A method of scanning in which all the lines in the frame are scanned out sequentially. Also known as progressive scan.

Normalized Correlation – Removes the absolute illumination value from a traditional correlation, making the algorithm less sensitive to light variations.

NTSC (National Television Standards Committee) – A 60 Hz standard for encoding color video signals. Used in North America, Japan, and most of South America.




O

Object – The 3D item to be imaged, gauged or inspected.

Object Features – Any characteristic that is descriptive of an image or region, and useful for distinguishing one from another. A feature may be any measurable item such as length, size, number of holes, surface texture amount or center of mass.

Object Plane – An imaginary plane at the object, which is focused by the optical system at the image plane on the sensor.

Oblique Illumination – A lighting direction at an angle which emphasizes object features by shadows produced. Oil Mist – An environmental contaminant which builds up on optical surfaces.

Opaqueness – Degree to which an object does not transmit light. Opening – An erosion followed by a dilation, it is the opposite of the closing morphological operator.

Optical Computing – Performing operations usually handled by electronic, serial computers with

optical or photonic circuits/elements in parallel at near the speed of light. Orientation – The angle or degree of difference between the object coordinate system major axis relative to a reference axis as defined in a 3D measurement space.




P

PAL (Phase Alternation Line) – A 50 Hz composite color video standard used in many parts of the world. Pantone Matching System (PMS) – A system of describing colors by assigning numbers. Parallax – The change in perspective of an object when viewed from two slightly different

positions. The object appears to shift position relative to its background, and also appears to

rotate slightly. Parallel Processor – A redundant hardware design using a number of processors so multiple pixels may be processed at the same time.

Parent – An object which wholly contains another object called a child. Passthrough – The continuous digitization and display of an image or video. Pattern Recognition – A process which identifies an object based on analysis of its features. Peak White – The highest point in the video waveform that the video level can reach and still stay

within spec. If the level does become greater than peak white, it is referred to as whiter-than­

white. Pedestal – An offset used to separate the active video from the blanking level. When a video system uses a pedestal, the black level is above the blanking level by a small amount. When a video system doesn’t use a pedestal, the black and blanking levels are the same.

Perceptron – The basic processing element used in neural networks. A simple analog circuit with weighted inputs and a nonlinear decision element such as a hard limiter, threshold logic or sigmoid nonlinearity.

Phase Adjust – A method of adjusting the color in an NTSC or PAL video signal.

Photodiode – A single photoelectric sensor element, either used stand-alone or a pixel site, part of a larger sensor array. Photometry – Measurement of light which is visible to the human eye (photopic response). Photopic Response – The color response of the eye's retinal cones.

Pinhole – A small, sharp edged hole, acts as a lens aperture which produces a soft edged image, is distortion free, with a wide field of view and large depth of field.

Pixel – An acronym for "picture element." The smallest distinguishable and resolvable area in an image. The discrete location of an individual photo-sensor in a solid state camera.

Pixel Clock – Used to divide the incoming horizontal video into pixels. The pixel clock has to be stable (with a very small amount of jitter) relative to the incoming video or the picture will not be stored correctly.

Pixel Counting – A simple technique for object identification representing the number of pixels contained within its boundaries.

Pixel Drop Out – Timing problems in the video system can cause pixel drop out, which creates image artifacts.

Polarized Light – Light which has had the vibrations of the electric or magnetic field vector typically restricted to a single direction, in a plane perpendicular to its direction of travel. It is created by a type of filter which absorbs one of the two perpendicular light rays. Crossing polarizers theoretically blocks all light transmission.

Polarizer – An optical device which converts natural or unpolarized light into polarized light by selective absorption of rays in one direction, and passing of rays perpendicular to the polarizing medium. Usually fabricated from stretched plastic sheets with oriented, parallel birefringent crystals. The first polarizers were constructed with parallel wires.

Positioning Equipment – Used to bring the part into the field of view, or to translate when multiple images or views are required.

Precision – The degree of spread or deviation between each measurement of the same part or feature. Repeatability.

Presence Verification – To look for an object to ensure that it exists.

Prism – An optical device with two or more non-parallel, polished faces from which light is either reflected or refracted. Often used to redirect light as in binoculars.

Processing Speed – A measure of the time used by an imaging system to receive, analyze and interpret image information. Often expressed in parts per minute.

Progressive Scan – A method of scanning in which all the lines in the frame are scanned out sequentially. Also known as interlaced.

Profile – The 3D contour of an object.




R

Radiometry – Measurement of light within the entire optical spectrum.

RAM (Random Access Memory) – Computer memory that allows data to be read or written at a particular location without having to pass sequentially through preceding locations.

Random Access – The ability to read out chosen lines or windows of information from an imager as needed, without following the RS-170 standards.

Range Measurement – Determination of the distance from a sensor to the object.

Raster – A series of scan lines that make up a television or computer display. All of the scan lines that make up a frame of video form a raster. Raster Line – See Scan Line. Raster Scan – A scanning pattern, generally from left to right while progressing from top to

bottom of the imaging sensor or the display monitor. Generally comprised of two fields composed

of odd and even lines. Real-Time – In image processing, an operation or a function that is complete in one frame time is said to be performed in real-time.

Reflection – The process by which incident light leaves the surface from the same side as it is illuminated

Refraction – The bending of light rays as they pass from one medium (i.e. air) to another (i.e. glass), each with a different index of refraction.

Region – Area of an image. Also called a region of interest.

Registration – The closeness of the part to the actual position expected for image acquisition.

Repeatability – The ability of a system to reproduce or duplicate the same measurement. The

total range of variation of a dimension is called the 6-sigma repeatability. See Precision.

Resolution – A basic measurement of how much information is on the screen. Higher resolution means much more detail in the image. Resolution, Feature – The smallest object or feature in an image which may be sensed. Resolution, Image – The number of rows and columns of pixels in an image. Resolution, Pixel Grayscale – The number of resolvable shades of gray (i.e. 256). Resolution, Spatial – A direct function of pixel spacing. Pixel size relative to the image FOV is

key.

Reticle – An optical element with a pattern located in the image plane to assist in calibration, measurement or alignment of a system or instrument. Examples are cross lines or grids. RGB (Red, Green, Blue) – The three primary color system used for video color representation. Ringlight – A circular lamp or bundles of optical fibers arranged around the perimeter of an

objective lens to illuminate the object in the field below it. A wide variety of sizes are available on

both a stock and custom basis. Rotation – Translation of a part about its center axis from the expected orientation in X and Y space. Expressed in degrees.

RS-170 – The Electronic Industries Association (EIA) standard governing monochrome television

studio electrical signals. The broadcast standard of 30 complete images per second. RS-232-C – The Electronic Industries Association (EIA) standard governing serial communications over a twisted pair. Good to about 150 feet.

RS-330 – Standard governing color television studio electrical signals.

RS-422; RS-423; RS-449 – The Electronic Industries Association (EIA) standards for serial communication protocols intended to gradually replace the widely used RS-232-C standard.

Run Length Encoding – A data reduction method to code a binary image. For each line in an image, data is stored denoting only the starting location of a blob and object and the length of the run of that line over the object.




S

Sample Rate – How often the A/D converter will take a sample of the video. Determined by the pixel clock. Saturation – The amount of color present. One of the three properties of color perception along

with hue and intensity. See HSI.

Scaling – A method of altering the size of an image in the x and/or y dimensions, which changes the effective resolution of the image. Scan Line – an individual sweep across the face of the display by the electron beam that makes

the picture. Scanner (galvo & polygon mirror) – An image sensor which uses a swept or scanned beam of

light (usually a laser) to generate or acquire a one or two dimensional grayscale reflectance pattern. Scattering – Redirection of light reflecting off a surface or through an object. See Diffuse. Scene – The object and a background in its simplest form. A portion of space imaged for

investigation or measurement. Scene Analysis – Performing image processing and pattern recognition on an entire image. Segmentation – The process of dividing a scene into a number of individual objects or

contiguous regions, differentiating them from each other and the image background.

Serration Pulses – In a composite video system, serration pulses tell the receiver where the vertical sync is. Shading – The variation of the brightness or relative illumination over the surface of an object,

often caused by color variations or surface curvature. Shape – An object characteristic, often referring to its spatial contour. Shape from Shading – A 3D technique that uses shadows from interaction of the object and the

light source to determine shape.

Sharpening – An image processing operation which enhances edges. An unsharp mask adds a low pass filtered image to the original, resulting in edge enhancement. Shutter – An electrical or mechanical device used to control the amount of time the imaging

surface is exposed to light. Often used to stop blur from moving objects.

Signal-to-Noise Ratio – A basic measurement of image quality. The magnitude of the signal (what is desirable) divided by the amount of interference with the signal (noise). Usually described in decibels (dB).

Silhouette – A black and white image of an object illuminated by backlighting.

Simple Lens – A lens with only a single element.

Sinusoidal Projection – Use of a grating in which the dark stripes vary in their density sinusoidally across each one, rather than constant black. Improved profile or range discrimination is possible when used in a moiré type configuration.

Size – An object characteristic typically measured by x and y dimensions. Size may be expressed in pixels, the system calibrated units of measure or classes or size groups.

Sobel Transform – A 3x3 convolution used for edge enhancement and locating.

Solid-State Camera – A camera which uses a solid state integrated circuit chip to convert incident light or other radiation into an analog electrical signal.

Span – The allowance of gray level acceptance for thresholding, adjustable from black to white from 0 to 100%.

Spatial Light Modulator – A transparent screen used in optical computer systems to introduce an image into the optical processing path. Similar to liquid crystal computer display screens, their resolution approaches 512x512 and grayscale imaging 8 bits. Also SLM.

Spectral Analysis – Evaluation of the wavelength composition of object irradiance.

Spectral Characteristics – The unique combination of wavelengths of light radiated from a source or transmitter or reflected from an object.

Spectral Response – The characteristic of a sensor to respond to a distribution of light by wavelength in the electromagnetic spectrum.

Specular Reflection – Light rays that are highly redirected at or near the same angle of incidence to a surface. Observation at this angle allows the viewer to "see" the light source.

Speed – An object characteristic expressed in distance moved per unit time. Velocity. Image blur may be caused by high speeds unless strobes or shutters are used to "stop motion."

SRI Algorithms – A rich set of routines used for geometric analysis and identification developed at the Stanford Research Institute in the early 1970s. Four main steps are: 1) Convert the image to binary; 2) Perform connectivity analysis to identify each blob or object; 3) Calculate the core statistical features for image objects; and 4) Calculate additional user selected features.

Stadimetry – A range measuring technique based on the apparent size measurement of a known size object in the field-of-view.

Statistical (Theoretic) Pattern Recognition – Statistical analysis of object features to perform recognition and classification.

Stereo (Passive) – For imaging, the use of two cameras, offset by a known distance and angle, to image the same object and provide range, depth or 3D information. Active stereo uses a controlled or structured light source to provide 3D data.

Stereo Photogrammetry – See Shape from Shading.

Stereoscopic Approach – The use of triangulation between two or more image views from differing positions. Used to determine range or depth.

Strobe Duration – The amount of time, expressed in microseconds, during which the flash lamp (strobe) is at 90% intensity.

Strobed Light – Brief flashes of light for observing an object during a short interval of time, typically used to "stop" movement and resulting image blur. Strobes may use xenon flash tubes, banks of LEDs or a laser to illuminate the scene.

Structural (Syntactic) Pattern Recognition – Evaluation of the relationship of object features in a specific order, i.e. decision trees, to perform recognition and classification.

Structured Light – Points, lines, circles, sheets and other projected configurations used to directly determine shape and/or range information by observing their deformation as it intersects the object in a known geometric configuration.

Subpixel Resolution – Mathematical techniques used on gray scale images to resolve an edge location to less than one pixel. A one tenth pixel resolution is reasonable in the factory.

Syntactic PR – See Structural Pattern Recognition.

Sync – The portion of a video signal indicating either the end of a field or end of a line of video information.

Sync Generator – A circuit that provides sync signals.

Sync Pulse – Timing signals used to control the television scanning and display process. The horizontal sync triggers tracing of a new line from left to right, while the vertical sync initiates the start of a new field.

Synchronous – A camera characteristic denoting operation at a fixed frequency locked to the AC power line (typically 60 or 50Hz).




T

TDI (Time Delay Integration) Camera – Similar to a line scan camera, a TDI camera is comprised of a number of rows of pixels. As an object such as a web moves, the charge from one row is passed to the next row, synchronously continuing the integration. Requires far less illumination intensity than the standard line scan.

Template – An artificial model of an object or a region or feature within an object.

Template Matching – A form of correlation used to find out how well two images match.

Texture – The degree of smoothness of an object surface. Texture affects light reflection, and is made more visible by shadows formed by its vertical structures.

Thickness – The measurement in the third dimension (length and width being the other two) from one object surface to another using one or two 3D range sensors or other technique.

Thresholding – The process of converting gray scale image into a binary image. If the pixel's value is above the threshold, it is converted to white. If below the threshold, the pixel value is converted to black.

Throughput Rate – The maximum parts per minute inspection rate of a system.

Timebase Corrector – Repairs a video signal that has bad sync.

Top Hat – A morphological operator comprised of an opening followed by a subtraction of the output image from the original input image.

Transition – For an edge in a binary image, the location where pixels change between light and dark.

Translation – Movement in the X and/or Y direction from a known point.

Translucent – An object characteristic in which part of the incident light is reflected and part is transmitted. The transmitted light emerges from the object diffused.

Transmittance – The ratio of the radiant power transmitted by an optical element or object to the incident radiant power.

Transputer – A type of computer architecture with several CPUs connected in parallel.

Triangulation – A method of determining distance by forming a right triangle consisting of a light source, camera and the object. The distance or range can be calculated if the camera-to-light source distance and the incident to reflected beam angle are both known. Based on the Pythagorean relation.

True Color – The imaging system is representing the image on the display with enough bits and in the correct format so that there is no loss of color image detail.

Tube Type Camera – A camera in which the image is formed on a fluorescent screen, then read out sequentially in a raster scan type pattern by an electron beam for conversion to an analog voltage proportional to incoming light intensity.




U

Ultrasonic Imaging – Use of ultrasound waves as the imaging "illumination" source.

Ultrasound – Low frequency radiated acoustical waves just above human sound perception which are useful for penetration and "illumination" for inspection of solid objects.

Ultraviolet – The region of the electromagnetic spectrum adjacent to the visible spectrum, but of higher frequency (shorter wavelength) than blue ranging from 1 to 400 nm. UV A ranges from 320 to 400 nm while UV B falls between 280 and 320 nm.




Validation – A rigid set of tests to verify that a system performs as documented.

Variable Scan Input – Frame grabber capability to accept a variety of non RS-170 input formats from a variety of cameras. Allows operation above the 30 Hz limit. Vector Scope – Used to determine the color purity of an NTSC or PAL video system.

Vertical Scan Rate – See Frame Rate. Vertical Sync – The portion of a video signal indicating the end of a field of video information.

This sync pulse is used by video equipment in order to maintain field synchronization with the incoming video signal. VESA (Video Electronics Standards Association) – A 32-bit display or other hardware card. VGA (Video Graphics Array) – The IBM video display standard of 16 colors. Video – Visual information encoded in a specific bandwidth and frequency spectrum location

originally developed for television and radar imaging.

Video Quality – A video image that has a high signal-to-noise ratio and is free of any image artifacts has high image quality. Video Waveform – Made up of several parts that are all required to make up a video image that

can be accurately displayed.

Vidicon – A generic name for a camera tube of normal light sensitivity. It outputs an analog voltage stream corresponding to the intensity of the incoming light. Visible Light – The region of the electromagnetic spectrum in which the human retina is

sensitive, ranging from about 400 to 750 nm in wavelength. VSYNC – See Vertical Sync.


W

Wavelength – The distance covered by one cycle of a sinusoidally varying wave as it travels at

or near the speed of light. It is inversely proportional to frequency.

Well – A morphological operator comprised of a closing followed by a subtraction of the output image from the original input image. Window – A selected portion of an image or a narrow range of gray scale values. Windowing – Performing image processing operations only within a predefined window or area

in the image.




Xenon Strobe – A gas filled electronic discharge tube, useful for high speed, short duration illumination for inspection.

X-ray – A portion of the electromagnetic spectrum beyond the ultraviolet with higher frequency and shorter wavelengths. Able to penetrate solid objects for internal, non-destructive evaluation.




Z

Zoom Lens – A compound lens which remains in focus as the image size is varied continuously. May be motorized or manually operated.

No comments: