一种基于室外光照估计的数字图片伪造检测算法  

A Digital Picture Forgery Detection Algorithm Based on Outdoor Illumination Estimation

作　　者：杨雪薇 邢冠宇 YANG Xue-wei;XING Guan-yu(College of Computer Science,Sichuan University,Chengdu 610065;National Key Laboratory of Fundamental Science on Synthetic Vision,Sichuan University,Chengdu 610000)

机构地区：[1]四川大学计算机学院,成都610065 [2]四川大学视觉合成图形图像技术国防重点学科实验室,成都610065

出　　处：《现代计算机》2020年第3期69-72,78,共5页Modern Computer

摘　　要：数字图片伪造已经逐渐成为信息可信度的威胁之一,如何衡量图片信息的可信度变得越来越重要。在各种各样的伪造图片中,人脸照片的拼接伪造有着非常重要的影响。针对此类伪造图片,提出一种新颖的基于室外光照分析的图片伪造检测算法来进行鉴别。通过人脸三维建模、光线追踪加速算法等对人脸图片进行预处理,采用改进后的普利兹天空模型结合漫反射模型设计能量方程并对光照模型参数进行优化,进而计算人脸所处环境光照条件。通过判断每张人脸还原出的光照是否一致,得出图像的真伪结论。对于室外光照下的人脸图片,该方法更具有优势,能够准确地还原出拍摄人脸照片当时的光照环境。Digital image forgery has gradually become one of the threats of information credibility.How to measure the credibility of image information becomes more and more important.In a variety of forged pictures,the splicing of face photos has a very important influence.Aiming at such forged pictures,proposes a novel image forgery detection algorithm based on outdoor illumination analysis for identification.The face image is preprocessed by three-dimensional face modeling and ray tracing acceleration algorithm.The improved Preetham sky luminance model is combined with the diffuse reflection model to design the energy equation and optimize the illumination model parameters to calcu⁃late the environment of the face.Lighting conditions.By judging whether the illumination restored by each face is consistent,the true and false conclusion of the image is obtained.For face images in outdoor lighting,this method is more advantageous,and can accurately re⁃store the lighting environment at the time of photographing the face.

关 键 词：图像伪造 光照估计 三维重建 非线性优化 

分 类 号：TP309[自动化与计算机技术—计算机系统结构] TP391.41[自动化与计算机技术—计算机科学与技术]