Machine VisionOptics in ManufacturingBy: Dan EhlkeFebruary 25, 2009
What is Machine Vision ?Application of Computer vision to industry & manufacturingMachine Vision is a subfield of engineering that incorporates:computer scienceopticsmechanical engineeringindustrial automation
Machine Vision Applicationinspection of manufactured goods such as:semiconductor chipsautomobilesfood Pharmaceuticalsand more . . .
Machine Vision CapabilitiesChecks for:Gauging or measurement : DimensionsSerial numbersPresence of componentsPattern matchingBlob analysis or Edge detectionOptical character recognition (OCR) 	and barcode decodingSurface inspectionColour analysis
General Process 1. Image Capture 2. Image Preprocessing 3. Definition of one or more (manual) regions of interest 4. Segmentation of the objects 5. Computation of object features 6. Decision as to the correctness of the segmented objects
Typcial System1 or more digital cameras (CCD or CMOS sensor)Optics : Lenses / mirrorsLight SourcesInput / Output hardwareSynchronizing Trigger sensorActuatorA program
Image Capturenote the difference between a vector & bitmap image
Vector vs BitmapExample: .SVGLine vectors & coordinatesBezier curveExample: JPEGPixelsGrid of values
LightingFront LightingBest for surface itemsPrintEtched numbersBack LightingBetter contrast for:GaugingpositioningRing lightingIntense, shadow-free lighting along optical axisDirectionalPicking out surface effectsDiffusedBetter for positioningLess shadows & reflectionsPolarized lightWith polarized lens on cameraPolarizing back lighting can even render surface tensions visibleInfrared or UltravioletCamerasA video camera produces analog signal that is digitized by a frame grabbercaptures individual still imagesLinescan camera
Multiple CamerasPart handling is simplified - part moves through on conveyorPart orientation is not requiredInspection is done without slowing manufacturing lineOnly minimal spacing between parts is required
Multiple camerasProcessing of the image
Image ProcessingRegions of Interest (ROIs) limit the information being analyzed.
LensesEntocentric magnification changes, due to object displacement image distortion perspective errors poor image resolution object edge position uncertainty, due to lighting geometryTelecenricthe front lens must be at least as large as the object largest dimension Largerheavier more expensive than common optics.Perecentric
Telecentric LensMagnification constancyIn Measurement applications, an orthonormal view is commonly desired for correct measurement.
Telecentric LensPerspective Errors Limited
Telecentric Lensimage size remains almost unchanged when the object distance changesimage size remains almost unchanged when the object distance changes
Telecentric LensDistortionDepends upon distance of object as well as the opticsTrapezoidal DistortionDifficult to calibrate out because it is asymmetricalRadial DistorsionNo Distortion
Telecentric LensResolutionTelecentric Lens
Telecentric LensNo edge position uncertainty
Perecentric Lens3D peripheral vision of objects without any aid of mirrors. Just 1 camera makes the job of 4 or more cameras placed around and over the objectNo image matching, typical of “multi-point” imaging systems, is necessaryAll the orthogonal sides have the same magnification because the object is seen from its top, not from one sideSpace between the object and the lens is left in order to enable the passing of pieces in on line applications Almost any kind of illumination source can be integrated
The Processmaster 5G automatic inspection system is installed upstream of the kiln to inspect either green or biscuit tiles, preferably before and after glazing. Defective tiles are immediately identified and removed from the production process. Costs for glazing and firing of defective products are dramatically reduced and reject tiles are recycled back into the raw materials, further optimising the production process.
Resourceshttp://www.machinevision.co.ukhttp://www.mmjp.or.jp/ned/english/prod_lsc_first.htmlhttp://www.opto-engineering.com/telecentric.phphttp://www.surface-inspection.comhttp://en.wikipedia.org/
http://video.yahoo.com/watch/1677649/5641423http://www.youtube.com/watch?v=RhPD_EqSwCc&feature=related
Machine vision

Machine vision