Table of Content

31 March 2023, Volume 41 Issue 2

Previous Issue    Next Issue

Digital Media Forensics and Security

Traceable DNN Model Protection Based on Watermark Neural Network

LIU Yalei, HE Hongjie, CHEN Fan, LIU Zhuohua

2023, 41(2):  183-196.  doi:10.3969/j.issn.0255-8297.2023.02.001

Asbtract ( 2425 )   PDF (9105KB) ( 470 )  

References | Related Articles | Metrics

This paper proposes a multi-user traceability watermarking neural network approach to model security and copyright certification for deep neural networks (DNN). The watermark is generated by the key driver and embedded invisibly in the output images of the DNN model, hence realizing the intellectual property protection and copyright tracking of DNN model. A codec network is added to the DNN model to embed the watermark, and a two-stream tamper detection network is used as the discriminator. Thus, the problem of residual watermark in the output images of the model is solved, which, reduces the impact on the performance of DNN model and enhances the security. In addition, a two-stage training method is proposed in this paper to distribute different watermarked models to different users. When copyright disputes occur, another residual network can be used to extract the watermark image from the output image. Experiments show that the proposed method is efficient in distributing watermarked models, and is able to trace the source of DNN models embedded with similar watermarked images for multiple users.

Reversible Data Hiding in Encrypted Image Based on Dual-Domain Joint Coding and Secret Sharing

WENG Ke, QIN Jianhao, SONG Tianran, SHI Hui

2023, 41(2):  197-217.  doi:10.3969/j.issn.0255-8297.2023.02.002

Asbtract ( 2178 )   PDF (7394KB) ( 213 )  

References | Related Articles | Metrics

A reversible data hiding scheme in encrypted image (RDHEI) based on dualdomain joint coding and secret sharing is proposed to improve the security and embedding capacity. First, the image is predicted by median-edge detector (MED) and the optimal threshold l is calculated. Pixels with different prediction errors are divided into predictable and unpredictable pixels according to the threshold. Second, to enhance the embedding capacity, the dual-domain joint coding is adopted to compress the auxiliary information on the pixel domain and bit domain, respectively. Then, the original image is encrypted to generate multiple encrypted images using the secret sharing technique based on cipher feedback, where the auxiliary information and secret data from multiple parties are embedded into the encrypted images. Finally, the embedded secret data as well as the original image are perfectly recovered based on the extracted auxiliary information. Experimental results show that the algorithm significantly improves the embedding capacity and security.

Segmented Backdoor Defense Based on Local Gradient and Global Gradient Ascent

XIAO Xiaotong, DING Jianwei, ZHANG Qi

2023, 41(2):  218-227.  doi:10.3969/j.issn.0255-8297.2023.02.003

Asbtract ( 2359 )   PDF (1267KB) ( 344 )  

References | Related Articles | Metrics

Backdoor triggers tend to be hidden and are difficult to detect. To solve this problem, a segmented backdoor defense (SBD) method based on local and global gradient ascent is proposed. In the early stage of training, local gradient ascent is introduced to enlarge the difference between the average training loss of backdoor samples and clean samples. A small number of high-precision backdoor samples are isolated to facilitate backdoor forgetting in the later stage. In the backdoor forgetting stage, global gradient ascent is introduced to reduce the correlation between backdoor samples and target categories to achieve defense. Based on three benchmark datasets GTSRB, Cifar10 and MNIST, a large number of experiments are conducted on the WideResNet-16-1 model against six advanced backdoor attacks. It is shown that the proposed segmented backdoor defense method can reduce the success rate of most attacks to below 5%. Moreover, the proposed method can train a clean equivalent learning model on both backdoor dataset and clean dataset.

Image Privacy Protection Based on Cycle-Consistent Generative Adversarial Networks

XIE Yiyi, ZHANG Yushu, ZHAO Ruoyu, WEN Wenying, ZHOU Yuqian

2023, 41(2):  228-239.  doi:10.3969/j.issn.0255-8297.2023.02.004

Asbtract ( 2484 )   PDF (9135KB) ( 759 )  

References | Related Articles | Metrics

Social media and cloud computing have facilitated the distribution and storage of images. Meanwhile, concerns about image privacy have been raised. It is crucial to protect image privacy from privacy violation and illegal use. Motivated by this, an image privacy protection method based on cycle-consistent generative adversarial networks (CycleGAN) is proposed in this paper. Considering the usability in image privacy protection, the method first combines image segmentation with CycleGAN to select different segmentation coefficients to generate images with different degrees of privacy protection. Then reversible information hiding is used to embed information in the generated privacy preserving image, so as to prevent illegal users from reversing the image. Thus, a balance is achieved between image privacy protection and usability. The proposed method is trained and tested using PIPA dataset, using peak signal to noise ratio and structural similarity index are used as performance metrics to evaluate the privacy-preserving images. Experimental results show that the proposed method outperforms other schemes in both image privacy preservation and usability.

Research Progress on Glyph Perturbation for Anti-print Scanning and Anti-screen Shooting

WANG Chen, YAO Ye, LI Li

2023, 41(2):  240-251.  doi:10.3969/j.issn.0255-8297.2023.02.005

Asbtract ( 2202 )   PDF (3084KB) ( 249 )  

References | Related Articles | Metrics

In this paper, we provide a review on the research progress of glyph perturbation. Following an introduction of the application scenarios of glyph perturbation, the existing works related to glyph perturbation for anti-print scanning and anti-screen shooting are systematically presented. The existing methods can be classified into two categories: traditional glyph perturbation and deep learning-based glyph perturbation. According to the attributes of characters, the former can be subdivided into pixel flipping, height adjustment, spacing adjustment, stroke adjustment, feature point adjustment, and skeleton adjustment. From the perspective of high-dimensional features, the latter slightly modifies the glyph of characters to embed additional messages. According to the property of the generated data, it can be divided into perturbed text image generation and vector font generation. The glyph perturbation methods are compared and analyzed in terms of robustness, embedding capacity, and complexity. Finally, the challenges and prospects of glyph perturbation are summarized.

Communication Engineering

FBG Temperature Distribution Detection Based on TS-DFT High Speed Spectral Demodulation

YANG Mei, CHEN Na, LIU Zhenmin, SHANG Yana, LIU Shupeng, CHEN Zhenyi, WANG Tingyun

2023, 41(2):  252-261.  doi:10.3969/j.issn.0255-8297.2023.02.006

Asbtract ( 2249 )   PDF (2320KB) ( 245 )  

References | Related Articles | Metrics

In this paper, high speed demodulation of fiber Bragg grating (FBG) reflectivity spectrum is realized based on time-stretch dispersive Fourier transformation (TS-DFT). The demodulation system consists of mode-locked laser, optical circulator, dispersion compensating fiber, reference FBG, sensing FBG, and data collection and processing module. The time domain mapping spectra of sensing FBG under different temperature fields were obtained in experiments. By comparison with the measured spectra of the optical spectrum analyzer, the high-speed spectral demodulation ability of the system is verified, and the demodulation rate is 51.2 MHz. Combined with the spectral inversion algorithm, the temperature distribution along the axis of sensing FBG is obtained, and the spatial resolution is 200 mm. Therefore, temperature sensing with high speed and high spatial resolution is realized.

Protected Wildlife Monitoring System with Low Power Consumption Based on NB-IoT

ZHANG Ximin, ZHAN Haisheng, LIU Qiang, YUAN Zhanjun, ZHANG Jinbo

2023, 41(2):  262-271.  doi:10.3969/j.issn.0255-8297.2023.02.007

Asbtract ( 2203 )   PDF (10189KB) ( 381 )  

References | Related Articles | Metrics

A distributed protected wildlife tracking and monitoring system based on NBIoT is proposed to realize low power consumption, remote monitoring, customization and low cost. The key solutions including efficient power management, concurrent communication processing, and network transparent transmission are given. The system consists of trackers and a monitoring information system. Each tracker includes a NB-IoT communication and location module, a low-power microcontroller, a solar composite power supply component with lithium battery, and a high-efficiency power converter. The monitoring information system is composed of data communication equipment, server and monitoring terminals. Experiments show that the system can collect the position of wild animals remotely and display their real-time positions and activity trajectories on the WEB map. Customized statistical analysis of the data and reports can be generated. The average power consumption of the tracker is less than 50mW, and the average positioning error is less than 20 meters. The system has a large monitoring range and low cost, and has been applied in a wildlife protection station showing good performance.

Signal and Information Processing

Reversible Information Hiding Algorithm in Ciphertext Domain with Multiple Embedding Based on Block Classification

ZHANG Xiangyu, LI Fengyong, QIN Chuan

2023, 41(2):  272-283.  doi:10.3969/j.issn.0255-8297.2023.02.008

Asbtract ( 2306 )   PDF (1824KB) ( 333 )  

References | Related Articles | Metrics

Insufficient utilization of image blocks in the existing reversible information hiding methods results in a low embedding capacity of secret information. In order to address this issue, this paper proposes a multiple embedding reversible information hiding algorithm based on block classification. First, the original image is encrypted with a stream cipher, and the encrypted image is further divided into multiple non-overlapping blocks. Subsequently, the Most Significant Bit (MSB) adaptive prediction algorithm is used to predict the first pixel and other pixels in each block, which is marked as a usable block or an unusable block. Finally, the available blocks are reconstructed and embedded with secret data, and the non-available blocks are re-embedded with the Median-Edge Detector (MED) prediction algorithm to realize the embedding of secret information. When the receiver receives the secret image, the secret information is extracted by the data-hiding key, and the original image is restored by the encryption key. Experimental results demonstrate that the proposed method can significantly improve the embedding capacity of secret information embedding, while keeping a high image restoration quality. The overall performance is superior to existing methods in terms of both embedding capacity and image restoration quality.

Adaptive Range Digital Beam-Forming Method for Ka-band Automotive Synthetic Aperture Radar

WEN Jing, HUA Wei, WANG Hui, GONG Qingkun

2023, 41(2):  284-295.  doi:10.3969/j.issn.0255-8297.2023.02.009

Asbtract ( 2095 )   PDF (19146KB) ( 195 )  

References | Related Articles | Metrics

The directional deviation on the topographic relief of the image arising from the low height of the vehicle is nonnegligible during the verification of digital beam-forming (DBF) using Ka-band automotive synthetic aperture radar (Ka-SAR). Due to the inevitable errors of the DBF weight coefficients using scan on receive (SCORE) algorithm, the synthetic beam pattern would deviate from the ideal state and degrade the system performance. In this paper, an adaptive range DBF processing algorithm based on multi-channel SAR is proposed. The DBF weighting coefficients are adaptively generated by interference processing and phase extraction, which improves the receiving gain. The adaptive algorithm has the advantages of high precision, simple processing flow, low computational load and real-time implementation. Finally, the effectiveness of the algorithm is verified based on simulation and experimental data.

Identification Method for Vessel Interrupt Track Correlating Based on Fuzzy Membership Degree

CHEN Zhaotong, CHEN Jiangping, PAN Li

2023, 41(2):  296-310.  doi:10.3969/j.issn.0255-8297.2023.02.010

Asbtract ( 2377 )   PDF (2053KB) ( 452 )  

References | Related Articles | Metrics

In order to integrate the vessels’ tracks from different monitoring sources and form a unified maritime posture, a track correlation identification method based on track prediction and fuzzy membership evaluation is proposed. A polynomial fitting method is used to speculate the past and future tracks. Position, course, and speed are selected as the fuzzy factors. The membership degree of each fuzzy factor between the predicted track is calculated by ridge fuzzy membership function and weighted to obtain the membership of a single moment. A weighting function is constructed to calculate the comprehensive membership degree and finally the threshold is set to determine whether the two tracks are correlated. In simulation experiments, the precision and the recall rate of the proposed method is higher than 90% and 80% respectively, which outperforms the traditional method. In order to further investigate the applicability of the proposed method, the radar monitoring data of vessel tracks under different scenarios are simulated, including stable scenarios, speed change and course change scenarios during the interruption interval. Simulation results show that the proposed method provides an effective way to solve the problem of correlation identification of interrupted tracks in the cross environment of non-cooperative vessels.

A Mosaic Puzzle Camouflage Steganography with Image Block Rotation

LIU Zhaozhi, ZHAO Yan

2023, 41(2):  311-325.  doi:10.3969/j.issn.0255-8297.2023.02.011

Asbtract ( 2100 )   PDF (3356KB) ( 150 )  

References | Related Articles | Metrics

In order to improve the efficiency and robustness of coverless steganography, a mosaic puzzle camouflage steganography with image block rotation is proposed. Firstly, the sender reduces the image randomly selected using the key in the shared image library and adjusts its tone. Then a random rotation angle is added to each image block according to the secret information and secret key to encode the image. Finally, the encoded image is arranged based on the rules to generate the secret mosaic image. The receiver removes the random rotation according to the secret key and obtains the secret information by reading the rotation angle of the encoded image. Experimental results show that the proposed method outperforms other coverless steganography methods in terms of embedding capacity and robustness.

Computer Science and Applications

Experimental Evaluation of FSM Conformance Testing Based on Structure Coverage and State Identification

LIN Weiwei, ZENG Hongwei, MIAO Huaikou, WANG Xiaolin

2023, 41(2):  326-343.  doi:10.3969/j.issn.0255-8297.2023.02.012

Asbtract ( 2060 )   PDF (786KB) ( 447 )  

References | Related Articles | Metrics

In finite state machine (FSM) conformance testing, there are two widely used test generation techniques which are based on structure coverage and state identification respectively. Under the condition of scarce test resources, we often face the problem of weighing selection of different test methods. To the best of our knowledge, there is no comprehensive comparative study of these two test techniques so far. This paper presents the necessity of experimental evaluation of the two test methods, and conducts experiments based on 10 FSM empirical cases. The performance is evaluated in terms of test cost and fault coverage capability, so as to provide empirical suggestions for the selection and application of these two techniques in FSM conformance testing.

Stock Price Trend Prediction Based on Dual-Stream LSTM Neural Network

WU Feng, XIE Cong, JI Shaopei

2023, 41(2):  344-358.  doi:10.3969/j.issn.0255-8297.2023.02.013

Asbtract ( 2230 )   PDF (1900KB) ( 592 )  

References | Related Articles | Metrics

Previous research on stock price volatility prediction relies on analyzing shallow features of financial news datasets and ignores the structural relationship between words in financial news, resulting in poor prediction performance. Aiming at this problem, we propose a stock price trend prediction model (Sent2Vec-DLSTM) based on a dual-stream long short-term memory network (LSTM) neural network. A vector generation model of emotional words called Sent2Vec is first proposed based on financial stock news data set and Harvard IV-4 emotion dictionary training, which is then combined with dual-stream LSTM neural network (DLSTM). In the experiment, the historical data of the S&P 500 index and the financial articles obtained by crawling are used to predict the trend of the S&P 500 index. the VietStock news and stock price data from cophieu68 are then used to predict the trend of the VN index. The results show that Sent2Vec-DLSTM outperforms existing models in stock price trend prediction.

An Electronic Contract Sharing Scheme Based on Blockchain

ZHAO Haihong, YAO Zhongyuan, ZHU Weihua, ZHU Ziqiang, PAN Changfeng, SI Xueming

2023, 41(2):  359-368.  doi:10.3969/j.issn.0255-8297.2023.02.014

Asbtract ( 2253 )   PDF (725KB) ( 293 )  

References | Related Articles | Metrics

In order to solve the problems of data tampering or leakage in the storage and sharing of electronic contracts, an electronic contract sharing scheme based on blockchain is proposed. First, a proxy smart contract is constructed by combining the contract with the proxy re-encryption to replace the traditional proxy, and the secure sharing of electronic contract is decentralized. Inter planetary file system (IPFS) is then used to store the ciphertext of electronic contract, and the electronic contract index address is stored in the blockchain, which effectively alleviates the storage pressure of the blockchain. Finally, the security performance of the proposed scheme is analyzed.