Design And Research Of Image CAPTCHAs Based On Human Visual Perception

CAPTCHA is a security mechanism designed to differentiate between computers and humans,and is used to defend against malicious bot programs.Text-based and image-based CAPTCHAs are two of the most widely deployed schemes,but most of the existing CAPTCHA schemes are either too difficult for humans to solve or not safe enough since a lot of researchers have attacked them successfully and These CAPTCHAs also suffer from language-dependent as well as cannot be automatically generated.But CAPTCHA plays a very important role in network security,so it is urgent to explore new possible schemes of CAPTCHA to solve those problem.In this paper,through the research of human visual perception ability,we designed three kinds of visual perception image-based CAPTHCA scheme,which can effectively solve the problem that the CAPTCHA field is facing,and make the field of image-based CAPTCHA continue to move forward.Through the study we found that human visual perception in the image field has a huge advantage compared with the computer.What is more,human has a very accurate and sensitive perception of the content of the image,the conversion of the style,the change of the space texture,and the three-dimensional space in the plane.Based on the characteristics of human visual perception,this paper designs and reseaches the image-based CAPTCHA based on human visual perception.In this paper,we have made two contributions to the field of CAPTCHA.Firstly,we propose three kinds of new image-based CAPTCHA mechanism based on visual perception by studying the human visual perception ability,from the three aspects of image style painting,image texture and three-dimensional spatial composition.The CAPTCHA of the Neural-Style mechanism contains two forms of Grid scheme and Font scheme.They are generated new style images by convolution neural network.The challenge of Grid scheme is asking users to find out the right answer according to the topic and the style-picture the users choose is in the grid,and Font scheme needs user to find out the Chinese characters in the style-picture.De Rection is one CAPTCHA scheme that takes advantage of human ability on capturing deformed regions of an image in the case of contrast.CONSCHEME is CAPTCHA scheme that capitalizes on human ability of understanding the content in a three-dimensional space and counting the number of stacked cubes in different views.Secondly,we propose a set of image-based CAPTCHA design guidelines.Compared with existing CAPTCHA schemes,the three new schemes proposed in this paper have something in common with existing image-based CAPTCHA.We conducted preliminary experiments over more than 110 people to verify the usability and security of two schemes.By analyzing the characteristics of these three schemes proposed in this paper and comparing them with the existing image-based ones,we propose a set of new standards for the design of image-based CAPTCHA.