- Advanced Steganography and Watermarking Techniques
- Digital Media Forensic Detection
- Chaos-based Image/Signal Encryption
- EEG and Brain-Computer Interfaces
- Digital Holography and Microscopy
- Internet Traffic Analysis and Secure E-voting
- Functional Brain Connectivity Studies
- Advanced Image Processing Techniques
- Diverse Aspects of Tourism Research
- Random lasers and scattering media
- Advanced Optical Imaging Technologies
- Neuroscience and Neural Engineering
- Peer-to-Peer Network Technologies
- Image and Video Quality Assessment
- Digital Rights Management and Security
- Media, Gender, and Advertising
- Electrochemical sensors and biosensors
- Fuzzy Logic and Control Systems
- Sparse and Compressive Sensing Techniques
- Vehicle License Plate Recognition
- Image Processing Techniques and Applications
- ECG Monitoring and Analysis
- Diverticular Disease and Complications
- Optical Imaging and Spectroscopy Techniques
- User Authentication and Security Systems
South China Normal University
2011-2024
City University of Macau
2024
Chinese People's Armed Police Force
2021
Chinese People's Armed Police Force Engineering University
2017-2021
State Key Laboratory of Cryptology
2019
In this paper, a novel data-driven information hiding scheme called "generative steganography by sampling" (GSS) is proposed. Unlike in traditional modification-based steganography, our method the stego image directly sampled powerful generator: no explicit cover used. Both parties share secret key used for message embedding and extraction. The Jensen-Shannon divergence introduced as new criterion evaluating security of generative steganography. Based on these principles, we propose simple...
This study aims to explore the possibility of using a multilayer artificial neural network for classification between children with autism spectrum disorder (ASD) and typically developing (TD) based on short-time spontaneous hemodynamic fluctuations. Spontaneous fluctuations were collected by functional near-infrared spectroscopy setup from bilateral inferior frontal gyrus temporal cortex in 25 ASD 22 TD children. To perform feature extraction classification, called CGRNN was used which...
A method for optical image hiding and encryption in the Fresnel domain via completely means is proposed, which encodes original object into encrypted then embeds it host our modified Mach-Zehnder interferometer architecture. The not only provides phase shifts to record complex amplitude of final on CCD plane but also introduces reference path hide it. registered as interference patterns, resemble a diffraction pattern image, thus secure information imperceptible unauthorized receivers. can...
As an antecedent of place attachment, on-site experience plays a key role in obtaining customer loyalty the context hot-spring resort operation. However, few studies have addressed consequences terms its components and examined empirically relationship between attachment. This study was focused on ascertaining dimensions examining through A conceptualization is proposed, with three dimensions: environmental experience, self-related social interaction. The results revealed interesting...
Spontaneous hemodynamic fluctuations recorded by functional near-infrared spectroscopy (fNIRS) from bilateral temporal lobes were analyzed on 25 children with autism spectrum disorder (ASD) and 22 typically developing (TD) children.By frequency domain analysis, a new characteristic was uncovered that the power of low cerebral oscillation showed distinct pattern in ASD.More specifically, at 0.0200 Hz, oxygenated hemoglobin larger for TD than ASD, whereas band 0.0267-0.0333Hz, deoxygenated ASD...
The steganographic model based on game theory is improved. A hybrid steganography that satisfies the Kerckhoffs principle by combining strategy adaptation of and content in cover proposed. This used to design a high-security algorithm for an H.264 video. distortion function UNIWARD improved according video coding characteristics. generate bias under model, embedding probability each position calculated. new generated cost using it embed information residual DCT coefficient block. experiment...
A single exposure optical image watermarking framework based on deep learning (DL) is proposed in this paper, and original watermark information can be reconstructed from only single-frame watermarked hologram by using an end-to-end network with high-quality. First, the acquired our presented phase-shifted interferometry (PSOIW) frame, then all holograms corresponding images are constructed to train datasets for of conditional generative adversarial (cGAN), finally retrieved well trained...
In order to keep the data secret, various techniques have been implemented encrypt and decrypt secret data. Cryptography is committed security of content, i.e. it cannot be restored with a given ciphertext. Steganography hiding existence communication channel within stego. However, has difficult construct cipher (cypher) that simultaneously satisfy both content for secure communication. Inspired by Cardan grille, this paper presents new generative framework grille cipher. A digital cardan...
A novel framework of optical image hiding based on deep learning (DL) is proposed in this paper, and hidden information can be reconstructed from an interferogram by using end to network with high-quality. By the prior data between object image, a generative adversarial was trained so that it learn model, which resulting only needs transmitted recorded reconstruct image. Moreover, reconstruction process obtained without parameters inverse diffraction result will not affected phase shifts...
With the continuous improvement in accuracy of steganalysis based on convolutional neural networks (CNNs), network scale has shown explosive growth. Consequently, CNNs have a high demand for hardware resources and time-consuming training. To reduce number CNN parameters improve efficiency steganalysis, we propose lightweight called W-Net. The proposed W-Net first uses grouped convolution channel shuffling units to extract noise residuals, strengthen information exchange between groups,...
Among steganalysis techniques, detection against motion vector (MV) domain-based video steganography in High Efficiency Video Coding (HEVC) standard remains a hot and challenging issue. For the purpose of improving performance, this paper proposes feature based on optimality predicted MVs with dimension one. Firstly, we point out that prediction (MVP) unit (PU) encoded using Advanced Motion Vector Prediction (AMVP) technique satisfies local cover video. Secondly, analyze HEVC video, message...
To assess the image quality of anal fistula by RESOLVE DWI compared to SS-EPI DWI, and explore value for diagnosis classification fistula. Qualitative quantitative evaluation showed that group was superior than group. The accuracy sequence in internal orifice, branch fistula, higher can provide images DWI.
This paper introduces the model structure used in SVDD 2024 Challenge. The challenge has been introduced this year for first time. Singing voice deepfake detection (SVDD) which faces complexities due to informal speech intonations and varying rates. In paper, we propose XWSB system, achieved SOTA per-formance challenge. stands XLS-R, WavLM, SLS Blend, representing integration of these technologies purpose SVDD. Specifically, best performing XLS-R&SLS from ASVspoof DF dataset, applied WavLM...
Video blogs (or vlogs), which are gradually becoming the mainstream way for young people to record their travels during COVID-19 pandemic, convenient introducing a travel destination through selfie or first-perspective long shot. This study investigates how narrative perspectives influence advertising effects, and explores moderating effect of tourists' affective risk perceptions. Two experimental studies were conducted test proposed hypotheses. Study 1 relationship between perspectives,...
In a data sharing network where parties lack mutual trust, credible traitor tracing is indispensable to reliable delivery. Currently, the main obstacles for are deficiency of content-independent watermarking scheme and untrustworthiness third parties. view this, we propose T-Tracer --- an encoding-based scheme. With T-Tracer, each party delivery embeds its digital fingerprint into byte sequence shared by generating tailor-made symbol mapping table, whereby can handle any type data. A...
Conventional optical image encryption methods based on phase-shifting interferometry need at least two interferograms, and the storage or transmission of interferograms needs to occupy a lot resources. At same time, low quality reconstructed complex natural images has always been main limiting factor in application security. In this paper, high-quality object reconstruction method from one-dimensional compressed encrypted signal multi-network mixed learning is proposed. First, an...