您的位置：首页 > 论文页面
一种基于图像纹理和BP神经网络算法的恶意文件检测方法
发表时间：2017-05-31 浏览量：1901 下载量：552
| 全部作者： | 温冠超，胡玉鹏 |
| 作者单位： | 湖南大学信息科学与工程学院 |
| 摘 要： | 针对传统的恶意文件检测方法存在检测速度慢、多平台检测适应性弱等问题，提出一种基于文件图像纹理和BP神经网络算法的恶意文件检测方法。通过结合图像分析技术和恶意文件检测技术，将恶意文件转换成灰度图像，使用灰阶共生矩阵（gray-level co-occurrence matrix，GLCM）算法和通用搜索树（generalized search trees，GIST）算法提取纹理特征，并基于BP神经网络算法进行学习训练，从而快速地识别出不同平台的恶意文件。实现了恶意文件的图像纹理提取和原型系统检测，并对virusshare项目上大量的病毒样本进行检测，实验结果表明，基于上述方法的检测具有速度快、多平台检测适应性强、准确度较高的特点。 |
| 关 键 词： | 软件工程；云存储；恶意文件；图像纹理；BP神经网络；恶意文件检测 |
| Title： | A malicious files detection method based on image texture and BP neural network algorithm |
| Author： | WEN Guanchao, HU Yupeng |
| Organization： | College of Computer Science and Electronic Engineering, Hunan University |
| Abstract： | In order to solve the problem better of slow speed of malicious files detection and weak adaptability of multi-platform detection in traditional way, this paper proposes a malicious files detection method based on image texture and BP neural network algorithm. By combining the technology of image analysis and the malicious files detection, the malicious files are converted into gray-scale images. Gray-level co-occurrence matrix (GLCM) algorithm and generalized search trees (GIST) algorithm are used to extract the texture features, and BP neural network algorithm is used for learning and training to detect malicious files from different platforms rapidly. This paper realizes image texture extraction and detection prototype system of malicious files. Through the experimental analysis of a large number of virus samples from the virusshare project, the experimental results show that the method proposed has the characteristics of fast speed, high adaptability in multi-platform testing and high accuracy. |
| Key words： | software engineering; cloud storage; malicious file; image texture; BP neural network; malicious files detection |
| 发表期数： | 2017年5月第10期 |
| 引用格式： | 温冠超，胡玉鹏. 一种基于图像纹理和BP神经网络算法的恶意文件检测方法[J]. 中国科技论文在线精品论文，2017，10（10）：1094-1105. |
3
评论数 0请您登录
暂无评论
下载全文
多维论文