Jyun-Jie Wang

An embedding strategy for large payload using convolutional embedding codes

A matrix embedding (ME) code is a commonly used steganographic technique that used linear block codes to perform the embedding process. However, a lack of low-complexity maximum-likelihood decoding schemes in linear block codes limited the embedding efficiency for sufficiently large lengths. This paper proposes a novel and practical hiding algorithm for binary data based on convolutional codes. Compared to a matrix embedding algorithm that uses linear block codes, the method proposed in this study is appropriate for embedding a sufficiently long message into a cover object. The proposed method employs the Viterbi decoding algorithm for embedding to identify the coset leader of convolutional codes for large payloads. Experimental results show that the embedding efficiency of the proposed scheme using convolutional codes is substantially superior to that of the scheme using linear block codes.

A suboptimal embedding algorithm with low complexity for binary data hiding

A novel suboptimal hiding algorithm for binary data based on weight approximation embedding, WAE, is proposed. Given a specified embedding rate, this algorithm exhibits an advantage of efficient binary embedding with reduced embedding complexity. The suboptimal WAE algorithm performs an embedding procedure through a parity check matrix. The optimal embedding based on maximal likelihood algorithm aims to locate the coset leader to minimize the embedding distortion. On the contrary, the WAE algorithm looks for a target vector close to the coset leader in an efficiently iterative manner. Given an linear embedding code C(n, k), the embedding complexity using the optimal algorithm is O(2k), while the complexity in the suboptimal WAE is reduced to O(sk) where s is the average iterations.

PAPR reduction in OFDM systems based on modified PTS algorithm with non-disjoint partition

A modified PTS algorithm by partitioning an OFDM block into non-disjoint OFDM sub-blocks is presented in this paper for PAPR reduction of M-QAM OFDM signals. Since a 16-QAM constellation can be written as sum of two QPSK sets, a non-disjoint sub-block partition of the 16-QAM OFDM block is obtained by applying two different disjoint partitions on QPSK OFDM blocks. Compared to a disjoint sub-block partition in conventional PTS, numerical results show that the modified PTS with a non-disjoint partition achieves an improvement of PAPR reduction and BER performance for interleaved, adjacent, and random partitioning schemes.

A low complexity PTS technique for PAPR reduction in OFDM systems

One of the major drawbacks of orthogonal frequency division multiplexing (OFDM) is the high peak-to-average ratio (PAPR) of the transmitted signal. The PTS method divides the input data block into disjoint subblocks and recombines them by using weighting factors. The searching complexity of the original PTS method increases exponentially with the number of these subblocks and is extremely high for a larger number of subblocks. In this paper we present a modified type of PTS algorithm combining with the minimal trellis of linear block codes, which not only reduces the searching complexity but also provides extra error-correction property.

Data Hiding Technique by Ternary Hamming Codes

A novel steganographic algorithm for ternay data based on ZZW construction over F_3 is proposed to improve the embedding efficiency. Contrast to the conventional ZZW construction using binary Hamming codes in the second channel, the proposed method employs ternary Hamming codes to embed the message at the second channel. For the first channel, a low rate convolutional codes is presented to enhance the embedding efficiency of this modified ZZW construction. The ZZW construction over F_3 provides excellent embedding efficiency for ternary signal, and can be greatly extended to embedding algorithm for non-binary steganography.

A trellis-based informed embedding with linear codes for digital watermarking

This paper proposes a novel informed embedding method based on a modified trellis structure for a digital watermarking system. This algorithm are capable of embedding adaptive robust watermarking bits according to various length of linear block codes in a host image of size 512×512 pixels. Instead of using randomly generated reference vectors as arc labels, this algorithm employs the codewords of a linear block code to label the arcs in the trellis. There are two advantages to use linear block code as arc. The first is to provide a satisfactory space partition for each trellis section, and the second is to perform embedding algorithm featuring linear block codes for each trellis section. Moreover, the algorithm proposed here is capable of performing iteration so as to find a trade off between robustness and fidelity by use of some controllable parameters. We report the robustness and fidelity performance of this algorithm in Gaussian noise and JPEG compression. Moreover, the proposed trellis-based informed embedding has much less computation complexity, compared to other informed embedding methods.

Joint JPEG-block coding with expurgating trellis for wireless robust image transmission

This paper proposes a joint JPEG-block coding system (JJBC) based on BCJR decoding with expurgating trellis that maintains an acceptable image quality over AWGN and wireless channels. The joint design of JPEG compression and block codes in the transmitter has the advantage of avoiding error propagation and of low decoding complexity over the conventional tandem system with separate optimization of source coding and channel coding. The PSNR performance of the JJBC system is much better than that of the tandem system when the signal to noise ratio is low. Experimental results show that the proposed JJBC system achieves 6 dB (3 dB) reduction at PSNR=20dB, compared to the conventional tandem system, over wireless IEEE 802.11b (AWGN) channels.