Image Region Copy-move Forgery Detection Based On Image Moments

With the development of computer technology and the popular of editing software for digital image, anyone can easily tamper an image. Copy-move forgery is one of the tampering techniques which are easily used. As a result, copy-move forgery detection has already become the research hotspot in recent years.In copy-move forgery detection, feature extraction and matching are two important steps. The image moments (e.g., Zernike moments, Exponential Fourier moments) are common feature vectors. At present, there are two problems in copy-move forgery detection. First, most of the existing methods for detecting altered regions are too sensitive to the post-processing operations, such as JPEG Compression, additive noise or Gaussian Blurring, and so on. Second, it's important to improve efficiency of detection. In view of the above questions, the paper proposes the solutions as follows.1. Image region copy-move forgery detection based on Exponential-Fourier moments. For RGB images, most existing copy-move forgery detection algorithms turn them to grayscale. When a RGB image turned to gray level image, the information of the RGB image was partly loss. A detection algorithm based on Exponential-Fourier moments is proposed. It detects three channels' images independently. Then the algorithms merge three results. Moreover, a model for calculating the critical threshold value is proposed, and the experimental results show that the model is reasonable. Compared with the method using Radial Harmonic Fourier moments, the proposed method is more robust.2. A Fast Algorithm of Image Moments in Copy-move Forgery Detection. As the running time of most existing feature extraction algorithms are long, we propose a fast algorithm of image moments in copy-move forgery detection. And it can reduce the running time. The fast algorithm uses the property of computational formula. In the final discretization expression, we can divide it into two parts. The one is the fixed section, the other is the grey level of the image. Through reducing the compute times of the fixed section, we can increase efficiency of the algorithm.3. An Improved Block-based Algorithm of Copy-Move Forgery Detection. Usually, the classical block-based matching algorithm (CBMA) can't find all matched sub-blocks. So we propose an improved block-based matching algorithm (IBMA) to solve the problem. It makes use of the sum of feature vectors to lexicographic sort the new matrix that formed by feature vectors and their sum. Then every row of the matrix will search the following rows until the difference in the first column is larger than a threshold value. When an image was distorted by Gaussian noise, salt-pepper noise, or JPEG compression, experiment results show that the improved block-based matching algorithm is better than the classical block-based matching algorithm.