Xingang You

Generalized transfer component analysis for mismatched JPEG steganalysis

Most universal JPEG steganalysis approaches rely on the assumption that training and testing samples come from the same distribution. They fail when training set and testing set are mismatched. In this paper, we propose generalized transfer component analysis for mismatched JPEG steganalysis to derive new representations from original features for training and testing samples to correct the mismatches. We first apply domain alignment to transform source domain (training set) to an intermediate domain closer to target domain (testing set). Then a set of common transfer components are learnt across two domains by minimizing the distribution distance between them. In the space spanned by these transfer components, two domains manifest similar characteristics and preserve enough discrimination to different categories. Extensive experiments demonstrate our method performs well in mismatched JPEG steganalysis.

SOURCE CAMERA IDENTIFICATION USING SUPPORT VECTOR MACHINES

Source camera identification is an important branch of image forensics. This paper describes a novel method for determining image origin based on color filter array (CFA) interpolation coefficient estimation. To reduce the perturbations introduced by a double JPEG compression, a covariance matrix is used to estimate the CFA interpolation coefficients. The classifier incorporates a combination of one-class and multi-class support vector machines to identify camera models as well as outliers that are not in the training set. Classification experiments demonstrate that the method is both accurate and robust for double-compressed JPEG images.

Silhouette coefficient based approach on cell-phone classification for unknown source images

Cell-phones have become a necessary communication accessory in daily life. MMS (Multimedia Messaging Service) used by smart phones has caused higher requirement on mobile image manipulation. Classifying image source cell-phones has become a major issue in the cell-phone communication forensics. There are two ways usually used for tracing and identifying the source device: image characteristics and equipment fingerprint. Both of the above schemes require a set of images captured by known source cell-phones for training a classification model. To avoid using any prior knowledge in practical scenarios, a graph based approach was proposed to classify the source cell-phones. Though an acceptable result has been obtained, a problem of incomplete classification appears in the case that one image is classified wrong into a single subset. In this paper, a silhouette coefficient based algorithm is proposed for source cell-phone classification. The spectral clustering algorithm is adopted in graph partitioning and the silhouette coefficient is used to extract the optimal classification from all the possibilities of classification. Experimental results show the validity of the proposed method.

Double Compression Detection Based on Markov Model of the First Digits of DCT Coefficients

Double compression usually occurs after the image has gone through some kinds of tampering, so double compression detection is a basic mean to assess the authenticity of a given image. In this paper, we propose to model the distribution of the mode based first digits of DCT (Discrete Cosine Transform) coefficients using Markov transition probability matrix and utilize its stationary distribution as features for double compression detection. Experiment results show the effectiveness of the proposed method and comparison has been made to show the improvement by using this second order statistical model.

Electric network frequency estimation based on linear canonical transform for audio signal authentication

As electric network frequency is sometimes embedded in audio signals when the recording is carried out with the equipment connected to an electrical outlet, electric network frequency estimation is an important task in audio authenticity. After the theoretical analysis, a novel electric network frequency estimation method based on linear canonical transform is proposed from anti-multipath interference point of view. The experimental results demonstrate that this model performs well with high precision in complex noisy environment.

Recording Device Identification Based on Cepstral Mixed Features

The authenticity of the recording evidence is the foundation of legitimacy and relevance, which is the primary condition of recording evidence. With the springing up of private recording evidence, there is an urgent need for authenticity identification of recordings. That the evidence shall be from an accurate and legitimate source is a prerequisite for three elements. Recording equipment identification is the core content of sources of evidence. This article studies the characteristics of the recording device parameters, proposing three characteristic parameters of recording equipment such as the proportion of time-domain low roughness, etc. And combined with improved Mel Frequency Cepstrum Coefficient (MFCC) feature parameters characteristic parameters constitute a hybrid 92-dimensional. According to experimental analysis, with 10 different brands and models of recording device (including five different brands and models commonly used in voice recorder and five kinds of commonly used different brands and models of mobile phones), 60 young men and women, each of 10 different voice, the same type of equipment to record each 2, shows that mixed characteristic parameters can effectively characterize the characteristics of the recording equipment. Recognition rate increases by more than 6% compared with ordinary cepstrum.

Effects on Statistical Features for Image by Quantization

The changes of statistical features can be used to detect the existence of secret messages in images. In this paper, it is indicated that statistical distribution of the whole AC coefficients, such as Generalized Gaussian, Laplacian, Cauchy, etc. can be greatly affected by JPEG quantization process. Furthermore, a constraint condition concerned with quantization step is presented under which the coefficients of individual channel can be fit Laplacian distribution. Based on the constraint condition above, a new steganalytic method, which can achieve a high degree of detection reliability, is also proposed.