Tuesday, November 9, 2010

Start of Pattern Recognition technology solutions

Buy something online, enter your email address and credit card number. Simple. Then you come to the lockbox, Completely Automated Public Turing test to Tell Computers and Humans Apart. Here, the Web site tries to confirm that you are a human, not a robot to commit cyber crime. Dutifully copied downward points deformed and watery.

Incorrect. Appears the another captcha. Please try again. Also wrong.A third aparece.Empiezas captcha to rethink your purchase.


University at Buffalo Venu Govindaraju, pioneer, along with colleagues at the UB, machine hand human handwriting recognition software believes that this annoying problem of the 21st century has a decidedly old-fashioned solution: handwriting.


"" Here at UB Center for unified biometrics, we are the only ones that have been proposed and thoroughly studied ink captchas, "says Govindaraju."Our perspective is that human beings are good at reading, handwriting, machines are not.It is naturally to humans. But researchers team normally considered writing hand a desperate, case until someone comes and shows them that it isn't.


Govindaraju must know. research, he and his colleagues UB carried out in the early 1990s helped United States postal service established the first machines that can read addresses handwritten, a feat that a number of time - particularly in the industry - said simply not hacer.En 1996, years of research, research enabled UB the USPS can start reading the machine address handwritten, increase efficiency and savings to the Agency millions of dollars each year.


Govindaraju believes that similar success can happen with captchas. One of his doctorate at the UB students have graduated and was hired by Yahoo! on the basis of his work on the development of "simulated" handwritten captchas.


"We have developed a file that can be generated automatically as a number of different styles of writing hand as we want," says Govindaraju.


Research is based on the recognition of patterns, a subfield of learning in computer science dealing with developing systems based on the detection of patterns in the data.


Similar questions are being studied by Govindaraju and his colleagues at the UB in order to develop "intelligent room", technologies supported by HP Labs Innovation Research Award.


"Smart rooms" are indoor environments equipped with sensitive but discreet, devices such as cameras and microphones that can identify and track the movements and gestures of inhabitants for a wide range of applications, from providing supplementary supervision life assisted for the elderly or disabled persons, places of Office work and establishments selling retail security monitoring facilities.Finally, the objective is to extend the functionality of "intelligent room" to the largest stadiums, such as shopping malls, airports and other transportation facilities.


Biometry Cubs scientists are studying for "intelligent room" applications include gestures of the hand, as well as common biometric facial recognition, voice and progress.


"This, too, is recognition of patterns," Govindaraju says, "but instead of letters, here, we are trying to standardize gestures."


"Is the development of an alphabet gestures so that machines can be programmed for gesture reconocimiento.La idea is to manipulate objects on a monitor without technology," said.


Since its founding in 2003, puppies has attracted approximately 10 million dollars in federal and industry funding and has produced 17 graduados.El doctoral level Center advances machine learning and pattern recognition technologies to build new methods to customize devices that use Biometrics data systems engineering to security of the homeland and civiles.Desarrolla applications physicist, such as fingerprints, hand geometry and iris; behavior, such as signing, the voiceprint and gait; and chemical biometrics biometrics as the smell of DNA and body scanning.

0 comments:

Post a Comment

Popular Posts