| Image is an important way to express and convey information, and it has been widelyused in many areas of society. With the development of digital image processing technology,the functions of image editing and processing software are increasingly rich and perfect, sothat image tampering and forgery becomes easy without leaving obvious traces. Abusiveusing of image forgeries has been faced great safety problem, in various fields like scientificresearch, political, economic, military and judicial areas, etc. Image forgeries will bringincalculable losses to people’s life and huge negative influence to the society. The authenticityand integrity of images are experiencing the serious threat and challenge. Therefore, imagetampering detection is one of the key issues to be solved in maintaining information security.In this paper, the PCA-SIFT algorithm has been improved and made use of CFAalgorithm for the detection of copy-move forgery, splicing forgery, and the identification ofcomputer generated images and real images, and other issues in-depth study. The mainresearch content is as follows:(1) This paper concludes the forming mechanism of image manipulation though a largenumber of literatures and analyses the effectiveness of the existing image tampering detectionalgorithm by lot of experiments as well as the existing problems.(2) This paper proposes an algorithm for detecting image forgery based on improvedPCA-SIFT. The presented method first extracts features of an image by SIFT and then furtherdescribes the feature though PCA and then makes use of KNN to operate feature matching.The algorithm is able to achieve the effective detection of image copy-move forgery.(3) This paper presents a way of image forgery detection based on mean square error andlocal variance of CFA features. The algorithm can realize the identification of computergenerated images and real images and the valid detection of image splicing forgery.Experiment results verified the effectiveness of the algorithms of our paper. |
PDF Full Text Request