Enhanced Camera Source Identification Based On Image Texture And Double Filtering Mechanism

With the development of science and technology, the various digital mediaresources have enriched people’s life. Meanwhile, many new problems and challengeshave emerged. With the help of all kinds of powerful image processing softwares,people can process and edit the digital image easily. Therefore, the research of digitalimage forensics has an important significance.Digital image source camera identification is a kind of digital image forensicstechnology, which can identify the imaging equipment of the digital image. This paperanalyzes several typical source identification algorithms, and the mainly work can besummarized as follows:1. Because using the sky image with simple texture and high brightness to extractnoise pattern can get relatively better accuracy, so many previous scholars get noisepattern by the sky image. However, in our real life, there are many ordinary sceneimages but few sky images. The source identification algorithm is proposed in thispaper based on texture. In proposed method, the noise pattern can be extracted fromthe ordinary scene image instead of the sky image. Through the experimentalcomparison and analysis, the effectiveness of the proposed method has been verified.2. The double filtering pre-processed mechanism is proposed for source cameraidentification. Thanks to the limits of the filter itself, some camera fingerprints stillexist in the low frequency component after filtering process only once. For thepositive samples, it still has a relatively high correlation value after filtering processtwice; while for the negative samples, it’s almost impossible to have a high correlationvalue. The experimental results demonstrate that the double filtering pre-processedmechanism can improve the performance of source camera identification..