Research On The Object-based Image Content Tamper-ING Forensics Methods

In recent years,with the rapid development of multimedia technology,the way to obtain digital information is becoming more and more convenient.Especially,as a most commonly digital information,image has been spread in all aspects of people's social life,and its authenticity and integrity seriously affect people's daily life.However,some people with ulterior motives have maliciously tampered with images for some purpose,which has caused serious social negative impact.Therefore,the forensics of the authenticity and integrity of image content has become very important.In this paper,we mainly study the technique of image hash and the method of image Copy-Move forgery detection,and the main research works are as follows:An object-based image hash method is proposed.The proposed method includes hash generation algorithm,and tampering detection and tampering localization algorithm.In the proposed method,we first extract SIFT feature points from the image,and reduce the dimension of feature vector by using the random Gauss matrix to construct the local location features.Then we use the SLIC algorithm to segment the image,and generate structural feature by extracting brightness,color difference and position information from image blocks.Third,we calculate the statistical magnitude of image blocks according to row and column,respectively,to extract the statistical feature of the image.Finally,we use the location feature,structure feature and statistical feature;combine with the size of the image and the number of initial seed points of SLIC segmentation to generate intermediate hash of the image.The final image hash is formed through encryption and randomization the intermediate hash.Based on this hash code,we propose a structural feature matching algorithm based on the position of the superpixel block to realize image tampering detection and tampering localization.A large number of experimental results demonstrate that the proposed method has strong robustness to most content-preserved image processing operation and it is very sensitive to malicious tampering.The proposed method hassatisfactory accuracy rate for tampered region localization.Compared with previous methods,it has obvious advantages.An image Copy-move forgery detection method based on feature points have been proposed.The proposed method includes image feature extraction and matching,false matching removing,affine transformation matrix estimation and tampering localization.In the proposed method,first,the SIFT feature points of the testing image is extracted,and the matching algorithm that combine the traditional SIFT feature matching and improved SIFT feature matching is used to realize features matching.Second,the mismatching point pairs are removed according to the property presented by the plane vector formed using the matching points.Then,an affine transformation matrix is introduced to calculate the transformation parameters between original region and duplicated region.The RANSAC algorithm is used to prevent the existence of a few mismatched point pairs and improve the parameter accuracy of the affine transformation matrix.Finally,the duplicated regions can be located by combining the affine transformation matrix and pixel correlation coefficient.A large number of experimental results show that the proposed method has good detection performance.Compared with other feature point based detection algorithms,the proposed method has satisfactory detection accuracy and robustness.