Man-machine identification method and system

Abstract

The invention discloses a man-machine identification method and system. The method comprises: when a registration request of a user is received, providing registration verification information which the user is requested to orally state; receiving voice data of the user, and establishing a voiceprint characteristic model for storage; when a login request of the user is received, providing login verification information which the user is requested to orally state; receiving voice data of the user, and establishing a voiceprint characteristic model, and comparing with the voiceprint characteristic model during registration of the user; and according to a comparison result, determining whether the user himself logs in. The man-machine identification method and system can improve declassification difficulties in man-machine identification.

Description

Technical field [0001] The invention relates to the field of computers, in particular to a method and system for human-machine identification. Background technique [0002] With the popularity of the Internet, various network services have increasingly become part of people's daily lives, such as e-commerce, free e-mail services, free resource downloads, and so on. However, these services for human users are often attacked by illegal users and abused by some malicious computer programs. They occupy service resources, generate a large amount of network junk, affect users' network experience, and pose a great threat to the security of network services. [0003] The human-machine identification system is a fully automated open-type human-machine distinguishing Turing test (Completely Automated Public Turing test to tell computers and humans apart, CAPTCHA), which uses security measures for question-and-answer authentication to distinguish between computers and humans. The operating m...

AI Technical Summary

Problems solved by technology

[0009] Text CAPTCHA recognizes people and machines by distorting text or characters, which prevents malicious registration or login of computer programs to some extent, but with the development of character segmentation and Optical Character Recognition (OCR) technology, most The text CAPTCHA has been successfully cracked, and the simple character recognition problem can no longer stop the computer program, and the distorted text is also difficult to recognize, making the user experience very bad

[0010] Image CAPTCHA takes advantage of the differences between humans and machines in image classification, object recognition, and common understanding. It is usually independent of different languages ​​and does not require user text input. Although it is more difficult to crack than text CAPTCHA, these image CAPTCHAs require huge database support and cannot be large-scale. Scale generation, in addition, is vulnerable to machine learning algorithms, such as: Golle designed a SVM (Support Vector Machine, Support Vector Machine) classifier that combines color and texture features to classify cat and dog images, and obtain High correct rate of 82.7%, the success rate of cracking Asirra containing 12 pictures can reach 10.3%

[0011] Sound CAPTCHA uses the difference in speech recognition between humans and machines to distinguish between humans and machines, but sound CAPTCHAs are also vulnerable to machine learning algorithms

Images