A Blind Forensics Method for Image Tampering

Abstract

An image tampering blind evidence collection method based on Sobel edge detection and image block brightness features, which is characterized by including the following steps: convert the suspicious image to be detected into a grayscale image, and convolve the grayscale image I(i,j) Process, obtain the gradient image G(i,j) of the image, perform threshold segmentation on G(i,j) to obtain the gradient image G(i,j), and perform binarization processing on the gradient image G(i,j) The binary image W(i,j) is obtained, and the binary image W(i,j) is converted into 1 pixel. By judging the similarity of the two sub-image sets, the present invention divides the image set into blocks. Brightness mean sorting can effectively improve the efficiency of the forensic algorithm, and by comparing similar brightness values ​​of image blocks, the accuracy is further improved, and the efficiency of image forensics is further improved; the present invention solves the problem of cloning tampering caused by large scale scaling. The problem of inability to accurately detect tampered images, by normalizing the brightness values ​​of image blocks, further improves the detection results of cloned images with different brightness values.

Description

technical field [0001] The invention relates to the technical field of information security and information processing, in particular to a blind image tampering evidence collection method. Background technique [0002] Image passive forensics technology, also known as image blind forensics technology, is a research field that has developed rapidly in recent years. It only uses the characteristics of the image itself to identify its authenticity, without any prior knowledge such as digital signatures or digital watermarks, and does not need to add watermark generation programs to camera equipment, so it has the advantage of a wide range of applications; at the same time , since there is no need to embed any information in the image, it will not reduce the image quality. When natural images are generated, they will be affected by factors such as the hardware and software inside the imaging device or the statistical laws of the natural images themselves. The output images will...

AI Technical Summary

Problems solved by technology

Similar methods include the DWT-SVD method of Li et al., the DWT-DCT (QCD) method of Ghorbani et al., and Huang et al. proposed an improved method based on discrete cosine transform based on the algorithm of Fridrich et al. They introduced a threshold p to control the dimension of the feature vector of the image block. The above method can reduce the matching time by p% on average, but as the value of p increases, more and more feature dimensions are discarded, and the accuracy of the algorithm also decreased

Images