Study On Key Security Technology Of Content-based Web Image Search Engine

In the virus world of Internet, it is an urgent and challenging problem that how can weensure the safety of search engines. Based on the research of content-based image searchengine system V2.0by our group, a security subsystem of the search engine is developed. Amalicious URL detection method based on BM pattern matching is proposed. a securitymethod by dynamic monitoring the search engine is proposed. At last, the security module iscombined into the web search engine content-based image search engine V3.0. The mainresearch content and results are as follows:Malicious URLs would be downloaded by web image search, and this may causeunnecessary loses to users. So the malicious URL detection method based on BM patternmatching is proposed. This method let the URL source code match the virus function toconfirm whether the URL is safe or not. Web image search detect203malicious URLsthrough this method. By kaspersky scanning，we confirmed189URLs to be malicious URLs,accurate rate is93.1%, false positive rate is6.9%. The experimental results show that themalicious URL detection module provides security URLs for web image search engine.When searcher download URLs and pictures from the Internet，the host may be crushedvirus. So the method which based on HOOK API dynamic monitoring image search engine isproposed. This method is mainly to hook system API functions. When virus program call theAPI functions, it actually execute the function which is written by us and were hooked in thesystem. When we turn on searcher to collect images, if registry and catalog have some specialchanges, monitor system will record these changes and feedback to users and users can handlethose changes to protect the security of search engine.Base on these research results above, we design and developed search engine securitysystem which is an important part of search engine system V3.0. This subsystem adopt pattern matching technology、HOOK API technology and monitor technology. The experiment resultsshow that this subsystem improves the security performance of search engine, and it canprovide substantial security image data for content based image search engine. It also canprotect a degree of protection to search engine where the server’s security, has achieved theanticipated target.