Double Compression Jpeg Image Detection And Median Filter Detection Research

With the rapid development of computer network and multimedia technology, the emergence of all kind of image generation equipment and image processing software (such as Neoimage, Adobe Photoshop, CorelDraw, and so on)owned rich function. Tampering with malicious and forging of the image becomes more and more esaily, and the fidelity of between real images and processed images is improved continuously. For the editing and processing of images in a certain range, on the one hand, makes our life rich and colorful, on the other hand, takes plentiful trouble and obsession to people’s normal life. Especially, images tampered or forged inflow the field of political, legal and so on, which will bring the huge damage and threat to the whole country interest and social stability. Therefore, there is important practical significance to judge the authenticity of the image, is the present multimedia safety technology research hot issues. And JPEG image format is the most popular image format at present. In view of this, this thesis studies the methods of JPEG doubly compression image detection and JPEG image median filtering detection, the main work is as follows:1. Proposing a algorithm for detecting JPEG doubly compressed image, this method analyzes the characteristics of JPEG single compressed image and JPEG doubly compressed image DCT transform quantization coefficient of histogram. Using the difference between JPEG ompression image (including JPEG doubly compressed images and JPEG single compressed images) actual quantization step and the estimated JPEG compression images quantization step, and the esimated JPEG compression images quantization step itself, as detecting JPEG doubly compressed images differentiating feature, to detect the JPEG doubly compressed images whether compressed twice or not. A large number experiments show that this thesis proposed method compared exsited method improved the accuracy detected JPEG doubly compressed image whether compressed tiwce or not, and making the range of detection images more widely.2. Analyzing the JPEG images median filtering smooth operation, proposed JPEG images median filter detection algorithm. This method using the characteristics of the median filtering, to be detected JPEG image make median filtering with3×3, meanwile, calibrate the detection JPEG iamges, and set up a certain threshold τ,caculate the variance a of image block. If varivance a less than threshold τ, the image block not participate in next computing. If variance a greater than threshold τ,the image block take part in next computing. Finally calculating the difference between the corrected image and image filtering, statisticsing total number difference image pixel point grey value is zero value and the mean value of the number of the mean image and the corresponding difference image pixel point grey value is zero, as two-dimensional features for detecting median filtering, used to detect whether JPEG image made median filtering operation or not. A lot of experiments show that this thesis proposed JPEG image median filtering detection algorithm compared exsited method not only improved the derection accuracy, but also greatly redued the required time.