Graphic identification code generation method for man-machine identification

Abstract

The invention discloses a graphic identification code generation method for man-machine identification, belongs to the field of network security testing, and solves the problems that an existing verification code is easily and automatically identified and skipped by a robot trained based on a machine learning technology, and an existing picture confrontation sample generation mode is long in processing time and large in calculation scale. The method comprises the following steps: presetting a picture size, selecting a picture set based on the picture size, inputting the picture set and a classification result of the picture set into an adversarial sample generation system to generate disturbance, and adding the disturbance to a picture frame of the picture set to obtain an initialized picture adversarial sample set; based on the initialized picture confrontation sample set, optimizing interference by using a white-box training model, and generating a final picture confrontation sample, namely a picture interference frame; and adding the picture interference frame added to the to-be-interfered picture, and forming a picture identification code, namely a picture verification code, wherein the size of the to-be-interfered picture is the same as the size of the picture in the picture set. The method is used for generating the interference frame.

Description

technical field [0001] The invention discloses a pattern identification code generation method for man-machine identification, which is used for generating an interference frame and belongs to the field of network security testing. Background technique [0002] With the development of artificial intelligence technology, repetitive and mechanical human labor is gradually replaced by machines, and there are also scenes on the Internet where machines are used to replace people to achieve some personal purposes (ticket grabbing, shopping spikes, web crawlers, etc.) , Greatly disrupted market fairness and network order. In order to prevent machines from performing operations instead of humans, human-machine recognition technology has also begun to develop rapidly and has achieved good results. At present, the main human-machine recognition methods include logical operation, picture feature recognition, picture behavior recognition, etc., by adding interference (rotation, blurrin...

AI Technical Summary

Problems solved by technology

[0006] In view of the problems of the above research, the purpose of the present invention is to provide a method for generating graphic identification codes for man-machine identification, which solves the problem that existing verification codes are easily automatically recognized and skipped by robots trained based on machine learning technology, and existing Image confrontation sample generation method takes a long time to deal with and calculates large-scale problems

Images