基于小波高阶统计量的数字图像来源取证方法-包装工程.PDFVIP

第38 卷 第 17 期 包 装 工 程 2017 年9 月 PACKAGING ENGINEERING ·179 · 基于小波高阶统计量的数字图像来源取证方法 王志红 1 ，黄惠芬 1,2 （1.山东英才学院，济南 250104 ；2.山东大学，济南 250100 ） 摘要： 目的 为了识别图像来源于相机还是其他电子设备，提出一种基于小波高阶统计量的图像来源取 证方法。方法 利用生成途径对图像性质的影响和数理统计方法，通过对比不同来源的数字图像的生成 途径对它们各自性质的影响，从图像的多尺度分析角度，研究不同来源的图像在变换域所呈现的独有特 性，提取数字图像小波域的高阶统计特征，以此判断图像的来源。结果 实验结果表明，针对计算机软 件、数码相机和扫描仪这 3 类图像的平均测试正确率达到了 97.9% 。结论 该方法在确定图像来源上有 较高的效率和准确率。 关键词： 高阶统计；特征提取；小波；识别 中图分类号：TP391 文献标识码：A 文章编号： 1001-3563(2017)17-0179-04 Forensic Method of Digital Image Source Based on Wavelet Higher-order Statistics 1 1,2 WANG Zhi-hong , HUANG Hui-fen (1.Shandong Yingcai University, Jinan 250104, China; 2.Shandong University, Jinan 250100, China) ABSTRACT: The work aims to propose an image source forensic method based on the wavelet higher-order statistics, in order to identify whether the image is from a camera or other electronic devices. Based on the influence of the generation path on the image properties and the mathematical statistics, and by comparing the influence of the generation path of the digital images from different sources on their respective property, the unique properties of images from different sources in transform domain were studied from the perspective of multi-scale analysis of the image. The high-order statistical characteristics of the wavelet domain in the digital image were extracted to determine the source of the image. The expe- rimental results showed that the average test accuracy rate was up to 97.9% for the three kinds of images from the com- puter software, digital cameras and scanners. The proposed method has higher efficiency an