基于光照色度一致性检测图像篡改的取证方法

165 浏览量更新于2024-08-29 收藏 784KB PDF 举报

本文主要探讨了基于图像光源色度一致性（Illuminant Chromaticity Consistency）的图像篡改检测方法。随着数字图像在日常生活中广泛应用，尤其是网络媒体中的图片分享，图像篡改已成为一个严重的问题，其中常见的手法包括同一图像的复制粘贴（Copy-Paste）和缩放等。为了提高图像篡改检测的准确性，研究者们提出了一种新颖的图像篡改鉴证方案，利用RGB三通道中光源色度的一致性作为识别图像是否被修改的指标。首先，文章指出传统的图像取证方法可能无法有效检测到同源图像内的局部修改，特别是当篡改区域与原始内容在色彩上保持一致时。作者意识到，光线在不同物体上的反射和散射可能导致颜色信息的变化，这种现象在不同对象的色度一致性中体现出来。因此，他们设计了一种算法，通过比较图像中不同对象的色度特征，寻找在篡改后的图像中可能出现的色度不一致的模式。具体来说，该方法涉及以下步骤： 1. **图像预处理**：对输入图像进行去噪和分割，以便准确提取每个区域的色度信息。 2. **色度一致性分析**：计算不同对象的色度分布，比如CIELAB或HSV颜色空间中的色彩特征，确保它们在正常情况下应该保持一致。 3. **篡改检测**：对比修改前后的色度分布，如果发现明显差异，尤其是在同一区域内复制粘贴或缩放后，可能存在篡改痕迹。 4. **置信度评估**：通过统计学方法，如方差分析或机器学习模型，评估色度变化的显著性，以确定篡改的可能性和程度。实验结果显示，该方法在检测复制粘贴和缩放等篡改操作时表现出良好的性能，能够有效地区分真实图像和经过修改的图像。这表明，通过关注光源色度一致性，可以为图像篡改检测提供一种更加精细且有针对性的策略，有助于提高鉴证结果的可靠性。本文的研究为图像取证领域提供了一个新的视角，强调了光源色度一致性在图像篡改检测中的重要性，并展示了在实际应用中其潜在的有效性。未来，该技术有可能被集成到自动化的图像监控系统中，进一步提升网络安全和证据保全的能力。

Forensics of Image Tampering Based on the

Consistency of Illuminant Chromaticity

Huang Yan-li

, Niu Shao-zhang

, Zou Jian-cheng

†

, Zhou Lin-na

Beijing Key Lab of Intelligent Telecommunication Software and Multimedia, Beijing University of Posts and

Telecommunications, Beijing 100876, China

E-mail: Deneae@163.com, szniu@bupt.edu.cn

†

Institute of Image Processing & Pattern Recognition, North China University of Technology, Beijing 100144, China

E-mail: zjc@ncut.edu.cn

Beijing Institute of Electronic Technology Application, Beijing 100091, China

E-mail: zhoulinna@tsinghua.edu.cn

Abstract—For the tamper method of the same image copy-

paste, a new tampering forensics scheme is proposed with the

consistency of illuminant chromaticity in the three color

channels, RGB, as the identification indicator of the image.

The experimental results show that the detection way, which

use the chromaticity consistency of different objects in the

image as the features of image tampering forensics, works well,

for the modifications of copy-paste and scaling on the pasted

region in the same image.

I. INTRODUCTION

Digital images are powerful tool for communication in

the daily life. With the development of image processing

techniques, it is not difficult to manipulate images for the

public users. Therefore, the photographs about news events

are no longer trusted by the major Medias, such as

newspapers, magazines, websites and television and so on.

In response, forensic techniques have emerged to detect

geometric, scene properties or statistical inconsistencies to

identify the authenticity of images.

This paper describes a new forgery detection method and

chooses the illuminant chromaticity as the unique

identification indicator of an image

[1,2]

. Based on the

theories of dichromatic reflectance model

[3]

, inverse-

intensity chromaticity(IIC)space

[4]

and Hough space, the

highlight regions of objects firstly are determined in the

image, and then the chromaticity values can be estimated

about three color channels. Next, the difference model of the

chromaticity between different objects is built, and finally

the difference value is compared with threshold value. If the

difference of the illuminant chromaticity exceeds the

threshold, the consistency of illuminant chromaticity should

be broken. Hence, the image can be authenticated as

doctored by this way.

II. D

ICHROMATIC REFERENCE MODEL

The dichromatic reflectance model states that the

reflection light happens when the light illuminates the

surface of non-uniform material. Diffuse reflection and

specular reflection consist of the reflection light on the

surface of objects in the image. Commonly, it assumes that

the chromaticity of specular reflection is close to the

illuminant chromaticity and camera response is linear. In a

camera, sensor response function is defined as (1):

() () () () ()

xxxxx

cscdc

GwBwI +=

(1)

Where

denotes one of the three color channels,

()

denotes the image intensity in the

color channel,

()

and

()

respectively the geometricparameters of

diffuse and specular reflection,

()

and

()

the sensor

response functions. The [4] states that the choice of

and

()

is not related with the location

, and shows that the

color of the specular light is globally constancy. Thus, the

image chromaticity function is (2):

()

∑

(2)

Where

()

is the image chromaticity in the point

and i belongs to one of the three color channels.

()

∑

(3)

()

∑

(4)

()

and

()

denote the chromaticity of diffuse and

specular reflection separately. So, the eq. 1 can be expressed

as follows:

() () () () ()

xxxxx

cscdc

ΓmΛmI +=

(5)

Where

() () ()

xxx

∑

idd

Bwm

(6)

() () ()

xxx

∑

iss

Gwm

(7)

APSIPA 2014

下载后可阅读完整内容，剩余3页未读，立即下载

weixin_38640443

粉丝: 10
资源: 885

基于光照色度一致性检测图像篡改的取证方法

The Art of Memory Forensics 无水印pdf

The Art of Memory Forensic

On the Security of Image Manipulation Forensics

NingBo-Univeristy.zip_forensics_image forensics_图像取证

A Robust Recoverable Algorithm used for Digital Speech Forensics based on DCT

The Art of Memory Forensics

A survey of passive technology for digital image forensics

The Art of Memory Forensics 内存取证

Forensics Analyzing an Unknown Image NTFS

Can we trust digital image forensics

最新资源