Research In Authenticity Of Digital Images Based On The Digital Featurs

Early the image’s authenticity is identified by checking the integrity of the specific digital watermark or digital signature which is in advance added in the image. Because of the need of image preprocessing, this technique once called digital image initiative forensic technique is limited in certain practical applications. With the development of forensic technology, there has been a new technique which known as digital image passive forensic technology to identify the authenticity of digital images without verifying the digital signature, digital watermark and any per-embedded information.Now searching for a new technique, which will be close to life, more convenient and helpful, become research emphasis. By studying and analyzing the implementation mechanism of existing identification techniques at home and abroad, research is focusing on numerical characteristics of the digital image, including to find the usable features by experiment, and use these features as a basis for the authenticity identification of digital images, and think about how to improve the forensic algorithm and make the range of application broader. This research is based on digital characteristics of storge format and image content.In the matter of image storage, the main object of study is the JPEG which is the image format nowadays widely used in personal computer. In the digital image forming process, the digital camera will generate a number of image information which is then stored in the JPEG Header. When the digital image is modified by image processing software, the image information stored in the JPEG Header will be changed, so this phenomenon can be helpful to identify the authenticity of digital images. Using these digital features, combine the image information stored in the JPEG Header and the image content, and then estimate the image information and get the deviation value by comparing with the real value, finally find out the threshold after fitting those deviation values curve for the digital images which will be identified. So if the image’s deviation values is not within the scope of the threshold value, the image can be sure is not really.In the matter of image content, the main object of study is Chinese character. Because Chinese character has a regular shape, the features is easy to find out. Using those features finds a faction that describe the relationship between features and image content, then get the threshold after comparing same falsified images with the real one, and last the image’s authenticity can be identified whether the estimate value is within the scope of the threshold value or not.