J4 ›› 2009, Vol. 27 ›› Issue (03): 293-.

• 论文 • 上一篇    下一篇

基于光照方向不一致性的图像盲鉴别方法

吕颖达,申铉京,陈海鹏,王友卫   

  1. 吉林大学 计算机科学与技术学院,长春130012
  • 出版日期:2009-05-20 发布日期:2009-07-13
  • 通讯作者: 吕颖达(1983— ),女,河北廊坊人, 吉林大学硕士研究生,主要从事图像处理与模式识别、计算机网络安全研究 E-mail:lvyingda1983@163.com
  • 作者简介:吕颖达(1983— )|女|河北廊坊人| 吉林大学硕士研究生|主要从事图像处理与模式识别、计算机网络安全研究|(Tel)86-13596436142(E-mail)lvyingda1983@163.com;申铉京(1958— )|男|吉林和龙人| 吉林大学教授|博士生导师|主要从事图像处理与模式识别、多媒体技术、智能控制技术研究|(Tel)86-13134456926(E-mail)xjshen@jlu.edu.cn
  • 基金资助:

    吉林省科技发展计划基金资助项目(20050503)                  

Blind Identification for Digital Images Based on Inconsistency of Illuminant Direction

L Ying-da,SHEN Xuan-jing,CHEN Hai-peng,WANG You-wei   

  1. College of Computer Science and Technology, Jilin University, Changchun 130012, China
  • Online:2009-05-20 Published:2009-07-13

摘要:

  图像篡改破坏了自然图像的光照一致性,为了保证图像内容的安全性,提出一种基于光照方向一致性的图像盲鉴别方法鉴别图像真伪。该方法根据图像的实际强度与计算强度间的误差函数以及适合不同光源模式的约束条件,估计不同光源模式下图像的光照方向。通过分析图像中不同对象或对象与背景之间的光照方向是否一致,判断图像是否被篡改。实验结果表明,无限光源模式的数字图像盲鉴别方法的正确检测率可达到81.8%,从而证明了局部光源和复合光源模式的数字图像盲鉴别方法的有效性。

关键词: 图像篡改, 光照方向, 盲鉴别, 图像真实性

Abstract:

Image tampering will destroy the light coherence of natural images. In order to ensure the security of image content, this paper presents a blind identification method based on the consistency of illuminant direction to identify the image authenticity. In accordance with the error function between the measured intensity and calculated intensity, and the regularization terms for different source models, this method estimates the illuminant direction for images of different source models. Analyzing if the illumination directions between different objects and background in one image are consistent to determine whether the image has been tampered with. Experimental results show that the correct detection rates for blind identification technology for digital images based on infinite light source model has reached up to 81.8%. In addition, experiments also show the effectiveness of blind identification base on local light source and composite light sources.

Key words: image tampering, illuminant direction, blind identification, image authenticity

中图分类号: 

  • TP391