The Feature Selection And Application Research Of The Sliding Block Verification Code Man-machine Identification System

With the continuous development of the Internet,the company's network technology is becoming more and more prominent.While the Internet provides people with rich resources and fast life,the security of network environment is becoming more and more prominent.Order to ensure the safe operation of the network,the network of register,login,search and other validation is particularly important,Internet companies,government departments and then into the human,material and financial resources to prevent the illegal ACTS of violence,to enhance the security of a web site and the hacker's attack.Therefore,it is of great significance to improve the exploration and research of verification identification technology and technology and to protect network security.At present,the application domain of network verification code is divided into number and letter combination verification code,mobile text verification code,sliding matching verification code,video verification code,etc.In order to verify the identification process,the implementation of each link will affect the effectiveness of the verification results.Identification technology at present mainly relates to image processing technology,character segmentation,pattern classification,owing to the complexity and diversity of network,the use of the authentication code is difference,so the study of its verification technology has important practical significance.Sliding matching of classification based on the verification code verification code as a case study,first of all,the slide to visiting users browsing data information for a series of analysis,and the obtained after data preprocessing,to extract the 12 indicators.Secondly,based on the analysis of the importance of indexes,the main component analysis,the Logistic coefficient method and the three methods were adopted respectively,and the 12 characteristic indexes were ranked in three different categories.In order to make full use of data information and sort the results of robustness,the sorting results of three different combination is obtained with consistent degree of combination method characteristics after the sorting,USES the results of such combinations is relatively reasonable and scientific;Finally,with the results of the collation sequence,the combination characteristics are identified by the K neighbor classification algorithm under different K values.Finally,10 feature indicators were obtained.Further analysis results show that under the same K value,increase gradually with the increase of feature recognition rate,but will be stable,the same characteristic indexes under different K value has a certain degree of difference,but with the improvement of recognition rate,eventually reached 90%,and the last two indicators,although to a certain extent to improve the recognition rate is influential,but little impact on the improvement of recognition rate,through to the validation of the test data set,further verify the results,the analysis of these results to some extent to the company,such as government policy makers to provide some reference value.Finally,on the basis of summarizing the research,it points out the direction to be studied.