Study On JPEG Forensic

With the popularization of digital cameras,mobile phones and other imaging equipment,and the rapid development of digital image processing technologies,people can easily acquire and process digital images,bringing great convenience to people's lives and also bringing about some Security risks.For example,some ulterior motives use tampered images to spread rumors,fabricate false news,and even use forged digital images as forensic evidence.This has serious and adverse negative effects on individuals and society.Therefore,it is of great significance to verify the authenticity and integrity of digital images.As the first international digital image compression standard,JPEG can save image data with extremely high compression rate and high-quality visual effects and is widely used on the Internet and various digital devices.Therefore,based on JPEG image forensic research has great practical value.This paper focuses on two aspects of research on JPEG image forensics:1)JPEG header detection.The JPEG header file contains image metadata and parameter information used for JPEG compression.Using image processing software to tamper with JPEG images will change the corresponding header file information.Using JPEG headers for image authenticity detection is a method of image forensics.The existing related algorithms are too simple and can only discriminate the originality of the image.This paper subdivides the specific conditions in the JPEG header file,classifies the detected problems,and not only can identify the originality of the image,but also can infer concrete tamper operation by detecting specific information,and give a reasonable evaluation result to the image according to the operation intention.2)Identification of digital image modification software.The tamperers process images by using existing digital image processing software such as Photoshop,Meitu,and n EO.JPEG images are often used during processing,leaving JPEG compression marks due to different digital image processing software.JPEG compression method is different,there are some differences in the remaining compression marks.This paper uses machine learning methods to extract the image residual characteristics to reflect these differences,by designing multiple classifiers to identify the compressed image of the software and parameters,experiments show that the proposed method in the processing software that not only can distinguish the compressed image effectively,but also can accurately identify the parameters used in the compressed image.