Blind Detection Based On Copy-move Tampering Of The Same Image

With the popularity of digital cameras,the popularity of high-performance smart phones and the continuous.innovation of high-performance cameras,in daily life digital images as information carriers are more and more widely used.With the digital image processing editing software is also a large number,the image processing has brought great benefits,but anything is "double-edged sword",which also gives those ulterior motives to bring harm to society more easily method.At present,the network and the community is full of a large number of forged pictures in the confusion,it has been harm to people's vital interests.Therefore,the identification and research of image tampering is of great significance,it should also refer to the research agenda.The current image tamper detection algorithm is diverse,Although our research on this aspect has made great achievements,we are still in the initial stage,have many shortcomings and the algorithm performance is relatively simple' In this paper,mainly for digital image area copy paste tampering means blind detection research,regional copy-move tampering is a local tampering way.At present,the blind detection technology for image area copy-move tampering is mainly based on image block detection algorithm and feature point detection algorithm.In this paper,the previous algorithm is analyzed and the corresponding improved algorithm is put forward.The main work of the author are:1.This paper first introduces the research background and significance of digital image forensics technology,and analyzes the status quo of image tampering systematically.It focuses on the copy-move of the same image,and analyzes the image copy-move model and an overview of the existing copy-move tampering blind forensics technology detection algorithm,introduced in detail the existing more classic detection method.2.In the third chapter,based on the image block detection algorithm and the advantages and disadvantages of the existing image block algorithm,an image block matching method based on improved Hu moment and Zernike moments is proposed.The algorithm is based on the improved Hu moments and Zernike moments to characterize the eigenvectors.The algorithm first slides the image and extracts the image block eigenvectors.The image sub-block feature vector correlation is used to identify the image tampering and locate its position.The algorithm real-time Sex is improved,and the translation,rotation is better robustness.The experimental results show that the algorithm can effectively resist the rotation and translation of the tampered area.3.In the fourth chapter,the algorithm of image blind detection based on feature point detection algorithm,analysis of traditional SIFT algorithm defects,proposed based on the fusion Gaussian geometric invariant moments of improved SIFT feature point detection algorithm.The algorithm uses the improved SIFT algorithm to extract the image key points,assigns the main direction to the feature points,and then extracts the Gaussian geometric invariant moments of the key neighborhood window as the characteristic descriptors of the key points,and finally the matching of the feature descriptors.The algorithm uses Euclidean distance to match the feature points,and uses the adaptive Euclidean distance threshold and RANSAC combined algorithm to eliminate the mismatch pair to realize the recognition and location of the tampered region.The experimental results show that the algorithm can basically keep the number of feature points extracted by the image,and even reduce the number of extracted feature points,but it can increase the number of feature points to reduce the number of matching points,and can improve the feature point extraction time,because the feature description And the matching efficiency is also improved.The experimental results show that the algorithm has a very good robustness to the tampered region translation,scale scaling and rotation operation,and the detection precision is also high.