Research On Splicing Image Detection Using The Statistical Features

With the rapid development and popularization of computer technology,digital image resources sharing is increasingly universal.Digital cameras and colorful images have quietly entered people's daily life and work.At the same time,with the emergence of a large number of digital image editing tools,the processing and modification of images has become more and more simple.Enjoying the convenience of the image in the Internet era,people also suffered the serious harm with the tampered images.Therefore,it is practically significant to judge the integrity and authenticity of the digital image content information.At present,the existing algorithms are mainly based on the statistical information of the image content to identify the tampered images.While the location of the tampered area is mainly achieved by tampering with changes in the edge of regions.In order to reduce the computational complexity while ensuring the accuracy of the detection,the thesis conducts a comprehensive study on the internal characteristics of splicing tampering images,and designs an algorithm to extract the effective features of the images,as well as implements the accurate identification of the pampered images and the precise positioning of the tampered area.The work of this thesis includes the following sections:(1)First of all,illustrates the significance of this thesis,and narrates the domestic and overseas research status of image tampering detection.By sorting the existing detection methods and then introducing the feature extraction method and the support vector machine used in these algorithms,provides the theoretical basis and research route for the image splicing detection algorithm proposed in this thesis.(2)Aiming at the theoretical model of the existing digital image tampering detection,the article offers a tamper detection algorithm in light of statistical features of image content.The method is mainly based on the fact that splicing tampering operation can destroy the original statistical features of the images,that is,the correlation between the adjacent pixels and the texture changes.By extracting the explicit features of these two changes,namely,the run length feature and the gray level co-occurrence matrix feature,the two feature vectors are effectively classified by using support vector machine classifier.In the end,through analyzing and comparing the results of the experimental in a standard data set,it can be found that the algorithm can extract theeffective features of the images before and after the tampering,thereby improving the accuracy of detection about the algorithm.(3)In view of that most existing algorithms are only used to detect the authenticity of the images,and not able to locate the tampering area of the splicing tampered images,this paper presents a splicing tamper detection algorithm on the basis of the correlation among image color components.The method is mainly based on the fact that color digital images in the imaging process will be the key operation,that is,CFA interpolation,which will make both the images among the color components and components among the pixels correlated.On this basis,it can determine the authenticity of splicing tampered images and locate the area of splicing tampering by extracting the image CFA interpolation correlation feature.The experimental results of this survey are compared with those of the existing algorithms in the same data set and working environment,and it is concluded that the algorithm can not only accurately determine the authenticity of the images,but also accurately position the tampered area,so the algorithm has a strong practicality.