项目名称: 基于本福德定律提高半脆弱水印图像篡改位置检测率的研究
项目编号: No.61202415
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 计算机科学学科
项目作者: 赵希
作者单位: 天津科技大学
项目金额: 25万元
中文摘要: 半脆弱水印,因其能容忍图像经受非恶意操作的特点,在用于检测图像的真实性和完整性上有日趋重要的作用。目前多数半脆弱水印算法的研究主要侧重在改善水印嵌入及提取算法上以提高图像篡改位置检测率。但其仅使用了一个固定阈值,忽略了图像遭受非恶意操作的不定性,结果将影响图像检测的误检率和漏检率。我们前期工作发现应用本福德定律预先估测图像的压缩比率,并依此选择合适的检测阈值能有效提高检测率。据此提出假设图像检测系统能预先检测出图像经受的多种非恶意操作程度,并据此调节检测阈值,就能最大限度降低误检率和漏检率。本项目将利用本福德定律,通过实验得出不同半脆弱水印算法中,经受不同非恶意操作图像的最佳检测阈值,建立图像检测阈值库,然后据此设计自适应阈值算法用于检测图像。本项目的研究成果可用于大部分半脆弱水印算法,且能在不改变原有算法的基础上,有效提高图像篡改位置检测率,这对半脆弱水印图像认证系统的完善有重要的意义。
中文关键词: 半脆弱水印;图像认证;本福德定律;检测阈值;图像压缩
英文摘要: Semi-fragile watermarking has become increasingly important as it can be used to verify the facticity and integrity of images, while tolerating some non-malicious manipulations. At present, the majority of semi-fragile schemes emphasis on improving watermarks embedding and detection algorithms, in order to improve the detection rate of localizing tampered areas. However, these algorithms have only applied a predetermined threshold, and it neglected the unpredictable non-malicious manipulations, which affect the false and miss detection rates. In our previous work, we utilized the Benford's Law to estimate the unknown Quality Factor (QF) of the watermarked images, and then according the estimated QF, appropriate thresholds could be adapted. Hence, we propose a hypothesis that the false and miss detection rates could be reduced if applying adaptive threshold by estimating the unknown non-malicious manipulations. In this project, we utilized the Benford's Law to analyses different non-malicious manipulations in different semi-fragile watermarking algorithms, to build an adaptive threshold database. The semi-fragile authentication systems could then select the most suitable threshold from this database according to the estimated non-malicious manipulations. The advantage of this research project is that could be plu
英文关键词: Semi-fragile Watermarking;Image Authentication;Benford's Law;Detection Threshold;Image Compression