基于内容的Web图像搜索引擎安全关键技术研究

发布时间：2018-06-28 02:24

本文选题：Web图像搜索器 + 恶意URL检测　；参考：《西安科技大学》2012年硕士论文

【摘要】：在互联网的病毒世界里,如何才能保证搜索引擎的安全已经是一个亟待解决且具有挑战性的问题。本文在课题组开发的基于内容的Web图像搜索引擎V2.0的基础上，开发了一套搜索引擎安全子系统。提出基于BM模式匹配的恶意URL检测方法；提出动态监测搜索引擎安全的方法；最后完成Web搜索引擎与安全模块的融合，构建出基于内容的图像搜索引擎系统V3.0。主要研究内容如下： Web图像搜索器可能下载到恶意URLs给用户造成不必要的损失，针对这一问题，提出基于BM模式匹配的恶意URL检测方法。该方法通过把图像搜索引擎下载的URL源码和数据库中的病毒特征进行模式匹配，来确定此URL是否安全。通过该方法检测恶意URLs共203条，用Kaspersky扫描确认是恶意URL的有189个，准确率为93.1%，误报率为6.9%。实验结果表明，恶意URL检测模块给搜索引擎提供了安全的URLs。当搜索下载网络上的URLs和图片时，，搜索引擎所在的主机很有可能会中毒。针对以上问题，提出动态监测搜索引擎所在主机安全的方法。该方法主要是对系统API函数进行挂接，当病毒程序调用这些函数时，实际上执行的是挂接在系统中的安全监测函数。Web图像搜索器运行后，当被监测的注册表或者系统目录有特殊变化时，监测系统把这些变化记录下来反馈给用户，用户可对这些变化进行处理来保护搜索引擎的安全。基于上述研究结果，设计并开发了一套搜索引擎安全子系统，该子系统是基于内容的图像搜索引擎系统V3.0的重要组成部分。该子系统采用了模式匹配技术、HOOK API技术和监控技术。系统测试表明，该子系统提高了搜索引擎的安全性，能较好地为基于内容的图像搜索引擎提供大量安全的图像数据，能在一定程度上保护搜索引擎所在主机的安全性，达到了预期目标。
[Abstract]:In the virus world of the Internet, how to ensure the safety of search engines is an urgent and challenging problem. Based on the content-based Web image search engine V2.0 developed by our research group, a set of search engine security subsystem is developed in this paper. A malicious URL detection method based on BM pattern matching is proposed, and a dynamic monitoring method of search engine security is proposed. Finally, the integration of Web search engine and security module is completed, and a content-based image search engine system V3.0 is constructed. The main research contents are as follows: Web image searchers may download malicious URLs to cause unnecessary losses to users. Aiming at this problem, a malicious URL detection method based on BM pattern matching is proposed. In this method, the URL source code downloaded by the image search engine is matched with the virus feature in the database to determine whether the URL is safe or not. A total of 203 malicious URLs were detected by this method. 189 URLs were detected by Kaspersky scan, the accuracy rate was 93.1, and the false positive rate was 6.9%. Experimental results show that the malicious URL detection module provides the search engine with a secure URLs. When searching and downloading URLs and images on the network, the host of the search engine is likely to be poisoned. Aiming at the above problems, this paper puts forward a method to dynamically monitor the host security of search engine. When the virus program calls these functions, it actually executes the security monitoring function. When there are special changes in the registry or system directory monitored, the monitoring system records these changes to the user, and the user can handle these changes to protect the safety of the search engine. Based on the above research results, a search engine security subsystem is designed and developed, which is an important part of content-based image search engine system V3.0. This subsystem adopts pattern matching technology and monitor technology. The system test shows that the subsystem can improve the security of search engine, provide a large amount of secure image data for content-based image search engine, and protect the security of the host to a certain extent. The target was achieved.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP393.09;TP391.3

【参考文献】