Research On The Copy-Move Passive Forensics Of Digital Image

Digital image, as an important information carrier, can intuitively describe information content therefore, it is widely used in daily life, especially can be seen everywhere on the internet. However, with the rapid development of image processing technology, and digital image editing software being more powerful, even non-professionals can easily distort the image content, which expressed a serious threat to the image of the authenticity and originality of the information. With the image content security problems have become more serious, digital image forensics technology came into being. As an important branch of the image forensics technology, digital image passive forensics refers to direct detection of suspicious images with no Pre-embedded digital watermark or digital signature logo.Copying and pasting the digital image is the most common hidden tampering means. Digital image passive forensics is mainly divided into two categories:tamper detection of the same image and different image. This article will focus on copy and paste tamper detection algorithms of the same the image.First, it is a brief introduction in recent years of digital image forensics passive research methods, and the classification of image detection algorithm. Then for the same image copy and paste tampering, the author respectively proposed two forensic algorithms based on wavelet moment and Dimensionality reduction CSIFT and their main innovation ideas.Forensics algorithm based on wavelet moments:as for the common translation, rotation and scaling geometric transformations tampering means of image tamper area, wavelet moments algorithm can well reflect the target invariant image features, such as rotation, scaling and translation invariance, so the wavelet moments algorithm are applied into image copy and paste tamper detection. First, using variation level to set active contour models suspicious areas delineated image, and then extract the suspicious region wavelet moment vector, and finally use similarity measure function to determine the area of the image tampering.Forensics algorithm based on Dimensionality reduction CSIFT:Since classical SIFT algorithm is applicable to the detection of gray image tampering detection, the false match rate of color images is high which greatly limits the use of the algorithm. Therefore, this paper, based dimension reduction CSIFT forensics algorithm, firstly use color invariant gradient direction as the image feature, and then use the nearest neighbor matching algorithm to match the feature points, and low-dimensional circular descriptor operator instead since the original algorithm 128-dimensional feature descriptor is very computational. Experimental results show that the dimensionality reduction CSIFT descriptor SIFT algorithm can not only overcome the shortcomings of using color information, but also improves the image feature point matching rate, and ensure the correct detection rate.