The Study Of Image Forensics Technology And Its Related Issue

With the extensive application of computing technology and digital equipment, the imagebecomes a record of our lives will leave at any time which bears a variety of valuableinformation. Image forensics is an emerging research mainly includes five aspects which isimage enhancement, image tampering evidence, indirect image information extraction,intelligent retrieval based on image content,build image based on the information, in order toexplore a variety of direct and indirect information of image,provide convincing evidence foroccasions of justice or identification. Image forensics technology development is still in itsinfancy, the study of domestic and foreign researchers gradually put in more effort in thisdirection,in which image enhancement, image tampering forensics image indirect informationextraction is the key and difficult points in the image forensics problem.This paper first introduces the image forensics research and the current status of the research,and then conducted in-depth research on three key issues of image forensics.For image enhancement, we construct a dyadic non-subsamped Contourlet transform to theimage denoising, this method of signal processing with multiple subbands, more detailedprocessing of the image signal, and use of non-subsamped filter for direction decomposition,so the signal is translational invariance,solve the image distortion caused by traditionalContourlet due to the lack of translational invariance, while non-subsamped filter to reduce noiseinterference and data redundancy. This paper supported by dyadic non-subsamped Contourlettransform’s muti-resolution and multi-directional, according to the image’s energy distribution invarious directions and scales, adaptive to eliminate image noise, experimental results show that,compared with wavelet denoising and Contourlet denoising, the algorithm can get higher PSNRvalues, while preserving image edges and details, and effectively reduce the Gibbs distortion,improve the visual effect.For image tampering evidence,this paper aims to retouch operation for imagetampering,according to the way of tamper showed in the frequency domain, combined withefficient image signal processing algorithms, proposed a blind detection forensicsalgorithm.Target image is first transform into the non-subsampled contourlet domain based on anew contrastàtrous wavelet, then put it into homomorphic filtering in the contourlet domain,inorder to enhance the edge of the fuzzy operation, and then combined with the morphologicalfilter to corrosion non-tampered region to complete the detection of tampering. After theexperiment, the algorithm has a excellent recognition results for image which was retouchtampered. For indirect image information extraction,this paper combined with defocus rangingalgorithm,for local motion blur of moving objects in the imaging process, according to thecorrespondence of move speed of objects and images blur scale, the target distance and exposuretime and other camera parameters,proposed a multi-focus velocimetry algorithm of movingobjects based on local motion point spread function. The movement scale of moving targetblurred image get by the autocorrelation peak of the edge image,the direction of motion obtainedby the frequency domain Radon algorithm. After the experiment, the proposed algorithm forreal-time speed of the moving target has a good recognition effect.After done the above work, the article summarized image forensics technology, and proposedfurther research content and research direction.