基于小波包分解的JPEG图像统计特性研究  

Statistical Analysis of JPEG Images Based on the Wavelet Packet Decomposition

作　　者：李智[1] 杨晓元[1] 魏萍[1] 殷三元[1] 

机构地区：[1]武警工程学院电子技术系,西安710086

出　　处：《计算机工程与应用》2003年第35期41-44,共4页Computer Engineering and Applications

基　　金：国家863高技术研究发展计划项目(编号:863-301-3);部委科研基金资助

摘　　要：该文首先给出了小波包分解系数绝对值的线性预测模型,由此提出了基于小波包分解的图像频域统计特性研究方法。然后针对JPEG图像研究其频域统计特性与文件大小之间,以及信息隐藏对频域统计特性的影响与其文件大小之间的关系。实验结果表明,JPEG图像频域统计特性与其文件大小之间具有相关性;信息隐藏工具Jsteg4.1对JPEG图像频域统计特性的影响与其文件大小之间具有相关性,而且主要表现为小波包分解系数的均值产生了显著失真;信息隐藏工具F5r11对JPEG图像频域统计特性的影响与其文件大小之间则不具有相关性。This paper introduces the linear magnitude predictor of wavelet packet decomposition coefficients,from which the research method of statistical characteristics of images in the frequency domain based on wavelet packet decomposition is proposed.Then the relationship between the statistical characteristics in the frequency domain and the sizes of JPEG images,as well as the relationship between the influence of the information hiding on the statistical characteristics in the frequency domain and the sizes of JPEG images is studied.The experiments results indicate that,the correlation between the statistical characteristics in the frequency domain and the sizes of JPEG images exists;the influence of the steganography tool Jsteg4.1on the statistical characteristics of JPEG images in the frequency domain is correlated to the sizes of JPEG images,and expresses mostly as the notable distortion of the means of wavelet packet decomposition coefficients;the influence of the steganography tool F5r11on the statistical characteristics of JPEG images in the frequency domain isn't correlated to the sizes of JPEG images.

关 键 词：信息隐藏 隐秘分析 小波包分解 线性预测模型 统计特性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]