Bei-bei Liu

An improved lossless data hiding scheme based on image VQ-index residual value coding

Copyright protection and information security have become serious problems due to the ever growing amount of digital data over the Internet. Reversible data hiding is a special type of data hiding technique that guarantees not only the secret data but also the cover media can be reconstructed without any distortion. Traditional schemes are based on spatial, discrete cosine transformation (DCT) and discrete wavelet transformation (DWT) domains. Recently, some vector quantization (VQ) based reversible data hiding schemes have been proposed. This paper proposes an improved reversible data hiding scheme based on VQ-index residual value coding. Experimental results show that our scheme outperforms two recently proposed schemes, namely side-match vector quantization (SMVQ)-based data hiding and modified fast correlation vector quantization (MFCVQ)-based data hiding.

On classification of source cameras: A graph based approach

Many existing source camera classification methods involve either training a classifier or computing the reference pattern noise of a camera, which means a set of images of known origins have to be pre-acquired. However, such requirement can not always be satisfied in real-world forensic applications. In this work, we propose a graph based approach that requires no extra auxiliary images nor a prior knowledge about the constitution of the image set. By formulating the classification task as a graph partitioning problem, a set of images can be classified according to their source cameras in an entirely blind way, with the number of source cameras automatically estimated. Experimental results have verified the validity of the proposed approach.

Source camera identification from significant noise residual regions

This paper investigates the digital forensic problem of determining whether an image has been produced by a specific digital camera. We employ the binary hypothesis testing scheme to detect the presence of photo-response non-uniformity( PRNU) in the image. The main challenge of this scheme is the extremely weak amount of PRNU in the observed noise residual. We propose to extract from the noise residual the significant regions with higher signal quality and discard those regions heavily deteriorated by irrelevant noises. Experimental results demonstrate that the proposed algorithm can improve the identification performance in the sense of decreasing the false rejection rate, which is a critical measure in practical applications.

Pornographic Images Detection Based on CBIR and Skin Analysis

A novel two-stage scheme of pornographic image detection is proposed in this paper. Specifically, we first apply the content-based image retrieval technique to find out whether human are present in the images. Then a detailed skin color analysis is performed to affirm the presence of pornographic content in the images. Experimental results show that the proposed algorithm performs well and fast in detecting pornographic images.

Robust Image Hashing for Image Authentication Based on DCT-DWT Composite Domain

Image hashing function maps an image into a bit string which can be used to authenticate the originality of the image. In this paper, we propose a perceptual image hashing scheme that is secure and robust to visually insignificant changes but fragile enough to detect and precisely locate malicious attacks. The proposed image hashing method is based on the outlines of one-dimensional signals re-arranged from the 8times8 DCT blocks. The final image hash is obtained by applying binary quantization to the DWT coefficients of the obtained 1-D signals. Experimental results are presented to show the effectiveness of the proposed scheme.

Comments on "A Semi-blind Digital Watermarking Scheme Based on Singular Value Decomposition"

For original paper see J. Shich, D.Lou and M. Chang, ibid., vol.28, p.428-440, (2006). In a recent paper, a robust water-marking approach for hiding grayscale image watermark into digital images based on SVD (singular value decomposition) is proposed by Shich et al. This comment demonstrates that the high robustness of the proposed scheme against various attacks is a result of improper algorithm design. The extracted watermark is just determined by the reference watermark used during the extraction process no matter what the input suspected image is.

Audio forensic authentication based on MOCC between ENF and reference signals

This paper proposes a new audio authenticity detection algorithm based on the max offset for cross correlation (MOCC) between the extracted ENF (Electric Network Frequency) signal and the reference signal. We first extract the ENF signal from a query audio signal. And then we partition it into overlapping blocks for forgery detection. The MOCC between the extracted ENF and the reference signal is calculated block by block. We also introduce an enhancement scheme to improve the quality of the ENF signal before the calculation of the MOCC. Our proposed method can detect not only audio forgery but also the edited region and audio forgery type. The effectiveness of our method has been verified by experiments on digitally edited audio signals.

An Improved Algorithm for Camera Model Identification Using Inter-channel Demosaicking Traces

Most CFA (color filter array) interpolation-based digital image forensic methods characterize inter-pixel relationship with a linear model and use the estimated interpolation coefficients as features for image source camera identification. However, various CFA models and interpolation algorithms must be tried for coefficient estimation during the detection process in that the CFA pattern of an image is often unknown at the receivers end. This incurs high computational complexity. Instead of using inter-pixel correlations, Ho et al. proposed to use inter-channel demosaicking/color interpolation traces for identifying the source camera model of a test image. In this work, we propose an improved algorithm. We first extract two variance maps by estimating the variances of each component of the green-to-red and green-to-blue spectrum differences, respectively, and then take the shape and texture features of these two maps for camera model identification. Experimental results show that our method achieves better overall performance.

Audio forgery detection based on max offsets for cross correlation between ENF and reference signal

The electric network frequency (ENF) is likely to be embedded in audio signals when the electronic recording devices are connected to electric power lines. If an audio signal is edited, the embedded ENF will be altered inevitably. In order to assess audio authenticity, this paper proposes a new method based on the max offset for cross correlation between the extracted ENF and the reference signal. By comparing the max offsets on a block-by-block basis, we can determine whether the audio signal in question was digitally edited as well as the location at which the editing manipulation occurs. The validity and effectiveness of our method have been verified by experiments on both synthetic composite signals and real-world audio signals.

Colorization Based on Image Manifold Learning

This paper concerns the issue of adding color to a grayscale image automatically. A novel colorization approach is proposed based on locally linear embedding (LLE), a manifold learning method. The colorization procedure utilizes the similarities of the image manifold structures between the grayscale space and color space. Colorized results can be obtained by transferring the parameters learnt from the grayscale space to the color space. Experimental results are reported to validate the proposed method