基于块峭度系数分析的数码照片真伪鉴别法  

A Method to Discern the Authenticity of Digital Photos Based on the Kurtosis of Block Data

作　　者：任先明 路林吉[1] 

机构地区：[1]上海交通大学自动化系

出　　处：《微型电脑应用》2013年第12期53-55,共3页Microcomputer Applications

摘　　要：从数码相机成像技术以及数字图像处理技术的原理出发,提出了以数码照片像素数据间的相关性特征对象"峭度系数"为要素,通过计算、比对照片上各重叠分块"块峭度系数"的实际值和评估值之间的差异来检测数码照片被修改的区域,从而对其进行鉴别及划类。鉴别实验表明,应用该法对于鉴别拼接而成的数码照片简单、快速、效果显著。This paper focuses on a method to discern the authenticity of digital photos based on the analysis on uniformity of pixel data that is described by the statistics paramcter, ＂coefficient of kurtosis＂, on the basis of principle of photography and image processing. The author design an algorithm to compute the ＂coefficient of kurtosis＂ of the adjacent overlapping blocks and make a comparison to search which parts have been modified. Experiments show that the method is simple, fast, and of accuracy and low false alarm rate.

关 键 词：数码照片 像素数据相关性 块峭度系数 鉴别 

分 类 号：TB852.1[一般工业技术—摄影技术]