Encyclopedia of Computer Science and Technology

112 computer visionsense on the part of the user remain an important last lineof defense.Further ReadingAntivirus Software Buying Guide. PC World. Available online. URL:http://www.pcworld.idg.com.au/index.php/id;316975074.Accessed June 24, 2007.CERT Coordination Center. Available online. URL: http://www.cert.org. Accessed June 24, 2007.Gregory, Peter H. Computer Viruses for Dummies. Indianapolis:Wiley, 2004.Henderson, Harry. Computer Viruses. Detroit: Lucent Books/Thomson-Gale,2006.McAfee Corporation. Available online. URL: http://www.mcafee.com. Accessed June 24, 2007.Symantec Corporation. Available online. URL: http://www.symantec.com. Accessed June 24, 2007.computer visionIn the biological world, vision is the process of receivinglight signals from the environment through the eyes andoptic nerves, from which the brain can extract patternsthat contain useful information (such as recognizing foodor a potential predator). Computer vision (also known asmachine vision) is the analogous process by which lightis received by a sensor system (such as a digital camera).The light is then analyzed for meaningful patterns. Thus, arobot might be able to recognize the identity and positionsof various parts on an assembly line.Because computer vision involves pattern recognition, itis part of the discipline of artificial intelligence (see artificialintelligence and pattern recognition). The challengeis not in getting information about a visual scene fromthe camera and turning it into digital information (a grid ofpixels). Rather, it is the ability to recognize meaningful patternsin fragmented images, something human infants learnto do almost from birth when they encounter human faces.One way to approach the problem is to constrain thekinds of images the computer (or robot) has to deal with.If you can guarantee that a robot’s field of vision will containonly a few fixed objects (a hopper, perhaps, or a conveyerbelt) plus one or more distinctively shaped parts, itis relatively easy to program the dimensions of the possibleobjects into the vision system so that the robot can identifyobjects by comparing them with stored templates. However,if the robot encounters an object it isn’t prepared for, suchas a stray bit of packing material, it will be unable to identify(or properly deal) with the object.Vision is also complicated by the problem of parsingthree-dimensional objects in the visual field. Seen headon,the side of a cube appears to be a two-dimensionalsquare. Seen at an angle, it appears to be a three-dimensionalassemblage with some faces visible and some not.To interpret these and more complicated objects, the robotmight be programmed with rules that help it infer that anobject is really a cube, that all cubes have six equal sides,and so on. Another strategy is to give the robot more thanone “eye” so that images can be compared, much as humansdo unconsciously with binocular vision. Finally, the robotcan be given the ability to move its head and eyes in orderto find a viewpoint that yields more information about anambiguous object.Human infants, of course, are not born with a fullydeveloped understanding of the types of objects in theirworld. They are always learning new ways to distinguish,for example, a stuffed teddy bear from a live dog. Robotvision systems, too, can be programmed to learn (or at least,refine their ability to recognize objects). A statistical techniquecan be used to “sample” objects in the environmentand find which characteristics most reliably “predict” thetrue nature of an object. Characteristics can be resampledfrom different viewpoints to see which ones remain invariant(unchanged). For example, a cube will always have fouredges on each face. Another approach is to use a neural network,where the visual information is processed by a grid ofnodes that are reinforced to the extent they are successfulin identifying features (such as edges).ApplicationsComputer vision is a problem of great theoretical interestbecause it engages so many questions about perception,the ability to build models of the world, and the ability tolearn. The field also has considerable practical potential.Currently, most robots are fixed to stations on factory floorswhere they work with a limited number of objects (parts)in a highly constrained, stable environment. However “servicerobots” have been gradually developed to work in amuch less constrained environment (such as carrying suppliesdown hospital corridors or even serving as mobileassistants to astronauts in the weightless environment ofthe International Space Station). These robots would benefitgreatly by having robust vision systems so that they can, forexample, recognize individual human faces or detect potentiallydangerous situations.Of course computer vision systems find many applicationsbesides robotics. These include automatic quality controlor inventory management systems, advanced medicalimaging and computer-assisted surgery, as well as security,surveillance, and criminal investigation/forensics.Further ReadingDavies, E. R. Machine Vision: Theory, Algorithms, Practicalities. 3rded. San Francisco: Morgan Kaufmann, 2004.Henderson, Harry. Modern Robotics: Building Versatile Machines.New York: Chelsea House Publishers, 2006.Hornberg, Alexander, ed. Handbook of Machine Vision. Weinheim,Germany: Wiley-VCH, 2006.Machine Vision Online. Automated Imaging Association. Availableonline. URL: http://www.machinevisiononline.org/.Accessed June 24, 2007.Shapiro, Linda G., and George Stockman. Computer Vision. UpperSaddle River, N.J.: Prentice Hall, 2001.concurrent programmingTraditional computer programs do only one thing at a time.Execution begins at a specified point and proceeds accordingto decision statements or loops that control the processing.This means that a program generally cannot begin onestep until a previous step ends.