Next Article in Journal
Divergence Measures for Globular T-Spherical Fuzzy Sets with Application in Selecting Solar Energy Systems
Previous Article in Journal
On the Kaiser Class of a Jonsson Theory
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Symmetry
Volume 17
Issue 6
10.3390/sym17060871
symmetry-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Review

Bibliometric Analysis of Digital Watermarking Based on CiteSpace

by
Maofeng Weng
1,
Wei Qu
1,
Eryong Ma
1,
Mingkang Wu
2,*,
Yuxin Dong
2 and
Xu Xi
2,*
1
Northwest Engineering Corporation Limited, PowerChina, Xi’an 710065, China
2
School of Geography Science and Geomatics Engineering, Suzhou University of Science and Technology, Suzhou 215009, China
*
Authors to whom correspondence should be addressed.
Symmetry 2025, 17(6), 871; https://doi.org/10.3390/sym17060871
Submission received: 12 April 2025 / Revised: 28 May 2025 / Accepted: 30 May 2025 / Published: 3 June 2025
(This article belongs to the Section Computer)
Browse Figures
Versions Notes

Abstract

Symmetries and symmetry-breaking play significant roles in data security. Digital watermarking is widely employed in information security fields such as copyright protection and traceability. With the continuous advancement of technology, the research into and application of digital watermarking face numerous challenges. To gain a comprehensive understanding of the current research status and trends in the development of digital watermarking, this paper conducts a bibliometric analysis using the CiteSpace software, focusing on 8621 publications related to digital watermarking (watermark/watermarking) from the Web of Science (WOS) Core Collection database, spanning from 2004 to 2024. This study explores the research landscape and future trends in digital watermarking from various perspectives, including annual publication volume, keyword co-occurrence and burst detection, leading authors, research institutions, and publishing countries or regions. The results reveal a regional concentration of research efforts, with early research being primarily dominated by the United States, Taiwan, and South Korea, while recent years have seen a rapid rise in research from China and India. However, global academic collaboration remains relatively fragmented and lacks a well-integrated international research network. Keyword analysis indicates that research hotspots have expanded from traditional copyright protection to data integrity verification, multimedia watermarking, and the incorporation of intelligent technologies. Notably, the introduction of deep learning has propelled watermarking algorithms toward greater sophistication and intelligence. Using CiteSpace, this study is the first to systematically illustrate the dynamic evolution of digital watermarking research over the past 20 years, focusing on thematic trends and regional distributions. Unlike previous reviews that rely mainly on qualitative analyses, this study offers a quantitative and visualized perspective. These findings provide concrete references for the future development of more targeted research efforts.

1. Introduction

Electronic data can be easily duplicated, transmitted, and misused in our contemporary information technology era, raising a number of security issues [1,2,3,4]. For both digital media producers and consumers, acts of piracy, tampering, and unapproved distribution present serious dangers and lead to financial losses. To address these challenges, researchers have proposed a variety of data security and copyright protection techniques, including encryption techniques, access control mechanisms, digital signatures, and authentication protocols. Encryption technology can effectively prevent data from being stolen during transmission, but it cannot prevent illegal copying and dissemination by legitimate recipients; access control increases the restriction on resource usage rights, but it is difficult to trace the responsible subject once the data have been leaked, while digital signatures can verify the integrity of the data, they usually cannot be embedded in the content itself. In contrast, digital watermarking technology has been widely researched and applied in the fields of copyright protection, tracking and tracing, and the content authentication of data due to its advantages of invisibility, robustness, and traceability [5,6].
At first, digital watermarking was primarily used to preserve digital images’ copyright [7,8,9,10,11]. This method makes it possible to verify ownership by embedding imperceptible but detectable information into an image. In their groundbreaking work, Schyndel et al. formally introduced the term “digital watermarking” in 1994, which sparked a wave of scholarly interest and established it as a crucial area of study in information security. In 1995, Cox et al. proposed a spread-spectrum-based digital watermarking technique, embedding watermark information within the frequency spectrum of digital signals, thereby enhancing both its imperceptibility and robustness [12,13]. The first International Information Hiding Workshop, which took place in May 1996, further underscored the growing prominence of this research domain, propelling digital watermarking towards further advancements.
The main focus of current research in digital watermarking is on its applications across various kinds of data, functional expansion, and performance improvement. Performance improvement efforts concentrate on enhancing the core characteristics of almost all watermarking algorithms: their embedding efficiency, robustness, imperceptibility, sensitivity, and security. In terms of functionality, significant advancements have been made in the use of digital watermarking for usage tracking, data authentication, and copyright protection. Its applications have been expanded to include data provenance, digital transactions, and security control in response to changing market demands, which continuously expand its functional scope. In the meantime, a growing variety of data formats of higher quality and complexity have been launched due to the fast development of hardware and software platforms. Digital watermarking is increasingly being applied to a greater variety of data products in order to meet increasingly high security requirements. Digital watermarking has gradually expanded from its original application in image copyright protection to a wide range of data types such as audio, video, text, maps, medical images, remote sensing images, 3D models, and so on. The continued development and improvement of digital watermarking technology has been fueled by this constant adaptability to various data formats [13,14,15,16,17]. Based on the application requirements and embedding characteristics of watermarking, the different types of watermarking can be broadly categorized as follows: (1) Robust watermarking: Mainly used for copyright protection and ownership statement, its core goal is to ensure the watermark can still be reliably extracted after common processing or attacks. (2) Fragile watermarking: Mainly used for data authentication and tampering detection. It is particularly suitable for scenarios that requiring strict data integrity, such as medical images and legal forensics. (3) Reversible watermarking: This watermarking enables the complete restoration of the original data after watermark extraction. It is widely used in applications requiring high data fidelity, such as remote sensing images, vector maps, and medical images. (4) Blind watermarking: This does not rely on the original carrier data during watermark extraction, which means that the watermark information can be successfully extracted without the need for the original data, which is of stronger practicality and security.
As technology continues to progress, digital watermarking’s fundamental workings, capabilities, and performance have also changed. Scholars have started investigating deep learning- and blockchain-based digital watermarking methods. Deep learning makes more effective and discrete watermarking possible while boosting resilience against attacks by training neural network models to embed and extract watermark information, leading to ground-breaking advancements in the field [18,19]. Blockchain makes it easier to store copyright data on the chain, which improves digital watermarking techniques even more. The findings of watermark authentication are more dependable and trustworthy when these technologies are used [20,21,22,23,24,25,26,27,28]. Furthermore, digital watermarking has become a research hotspot due to the digital media industry’s explosive growth and the growing needs of society for secure information sharing and intellectual property protection. Numerous creative research avenues have also been generated by the rise of new technology and changing demands. With many academics performing review studies concentrating on fundamental concepts [29,30,31], applications [32,33,34], performance [35], and specialized data types [36,37,38], the amount of research on digital watermarking has been continuously growing. Although existing reviews have analyzed the research results of digital watermarking in depth, most rely on expert experience and subjective judgment to organize the literature. They generally lack a comprehensive, quantitative analysis of research trends and emerging frontiers.
Bibliometrics is a discipline that takes a literature system and its related media as its research object and uses mathematical and statistical methods to carry out a systematic and quantitative analysis of knowledge carriers. It mainly reveals the mode of the dissemination of scientific knowledge, the structure of research activities, and the influence of academic achievements through quantitative research on the number of documents, author distribution, keyword frequency, citation relationship, and other elements; helps researchers to evaluate and predict scientific research activities; and provides data support for knowledge management and decision-making. Commonly used measurement methods include citation frequency analysis, co-citation analysis, and author cooperation network analysis. The knowledge maps constructed through bibliometric analyses visually represent the complex relationships within academic research. They intuitively show the structural links among research topics, evolution pathways, and the chronological evolution of research hotspots so as to provide researchers with a more insightful knowledge discovery tool. CiteSpace, a visualization tool widely used for academic research, was developed by Dr. Chaomei Chen at Drexel University’s College of Computing & Informatics. This software is a popular tool in bibliometric analysis and data visualization because it makes it easier to identify and visualize research hotspots within a given domain [39]. CiteSpace has been widely used in research trend analyses and bibliometric studies since its launch [40,41,42,43,44].
This study introduces bibliometrics and knowledge mapping methods into the field of digital watermarking for the first time. Using the CiteSpace tool, we systematically analyze relevant studies from the past 20 years, focusing on their temporal evolution, research hotspots, and scientific research cooperation networks. A comprehensive knowledge map is constructed to reveal the field’s knowledge structure and developmental trends, providing both scientific and theoretical references for future research. This study aims to objectively present the overall state of research on digital watermarking and to provide novel research perspectives and methodological paths for exploring its frontiers and emerging hotspots.

2. Methodology

This study utilizes CiteSpace to analyze the scholarly literature on digital watermarking since its inception, examining publication volumes, keywords, author networks, contributing institutions, and journals. Through visualization techniques, it explores and presents the knowledge structure underlying these publications. The version of CiteSpace software used in this study is 6.3.R164-bit (Chaomei Chen, Philadelphia, PA, USA; download link: https://citespace.podia.com/, accessed on 15 February 2024).
Although the idea of digital watermarking is well defined, it is frequently thought of as a method that is related to but different from data concealing, which is a subset of steganography. Since digital watermarking and data hiding share the same basic mechanisms, functions, and algorithmic principles, despite their different application circumstances, we view data hiding as an essential component of digital watermarking in our study. In order to ensure the reliability and quality of the literature sources, we chose the Web of Science Core Collection database, which contains high-quality and authoritative academic records. The search was conducted within the SCI-EXPANDED and SSCI editions using the query Keyword = “watermarking/Watermark” AND “Document Type” = “Article”. The search time range was 1 January 2004 to 31 December 2024, and only articles published in English were included. In addition, duplicate records were eliminated using CiteSpace’s built-in data preprocessing function.

3. Results and Analysis

3.1. Publication Trend Analysis

Figure 1 shows the annual publication trend for the 8621 publications that were gathered for this study between 2004 and 2024. In general, the quantity of articles published has been increasing, with only small decreases in a few of the years. The publication trend can be roughly separated into two phases based on growth patterns:
  • Period of Steady Development (2004–2016): The field experienced a phase of steady growth, accompanied by fluctuations in annual publication counts. While the number of annual publications remained between 200 and 400, the average annual growth rate during this period was about 3.6%, with a small decline in some individual years. These fluctuations reflect a research landscape primarily focused on robustness enhancement and conventional transform-domain watermarking techniques. Even though significant progress was made, new approaches and lines of inquiry were still required to spur additional development in watermarking technologies.
  • Period of Rapid Growth (2017–2024): Since 2017, the number of papers published has entered a phase of rapid development, with an average annual growth rate of more than 6%. The annual number of publications exceeded 400 for the first time in 2017 and continued to rise, reaching a peak in 2023, where it totaled 663, an increase of about 66% compared with 2016. This pattern demonstrates that digital goods are becoming more and more commonplace globally, as well as that demand for digital copyright protection is rising. Increased interest from the public and private sectors has spurred research, which has been further driven by improvements in regulatory frameworks. Digital watermarking is anticipated to continue to be a major area of research and application in the future as it firmly establishes itself as an essential technique for data security and digital copyright protection.

3.2. Author Cooperation Network Analysis

Table 1 displays the top 10 most prolific authors in the field based on a graphical analysis of authorship from 2004 to 2024. With more than 40 published papers between them, Chang, Chin-Chen, Zhang, Xinpeng and Singh, Amit Kumar are the top three contributors. Notably, Chang, Chin-Chen, the most published author in this field, has consistently produced research since the beginning of studies on digital watermarking and has continued to make important contributions to academia. The author collaboration network in watermarking research from 2004 to 2024 is depicted in Figure 2. The size of the nodes in this network is proportional to the quantity of publications attributed to each author; the early, middle, and recent stages of this field are denoted by blue, yellow, and red, respectively. Stronger co-authorship relationships are indicated by thicker connecting lines, which show the strength of the collaboration. Key metrics are shown in the upper-left corner of the figure: N = 300; E = 209; and Density = 0.0047. According to these data, the network has 300 author nodes and 209 collaborative links, with an overall cooperation density of 0.0047. This suggests that the research environment is somewhat dispersed and that highly centralized collaborations have not yet developed. Further analysis reveals that highly productive authors tend to engage in collaborative research, suggesting the emergence of a relatively stable network within the watermarking field. From a temporal perspective, single-author papers were more prevalent in the early and middle stages, whereas multi-author collaborations have become increasingly dominant in recent years. This shift toward team-based research not only fosters academic exchange but also enhances the depth and breadth of research outcomes. International cooperation remains limited, and the cooperation network lacks deep regional integration, instead showing localized clustering based on nations or regions. Reasons for the low density of collaboration may include the lack of stable cooperation mechanisms among researchers, language and cultural barriers, differences in research resources and policies in different countries, and the fact that some prolific authors prefer their own country or institution’s collaboration network.

3.3. Institution Cooperation Network Analysis

As seen in Figure 3, the institutional collaboration network offers a clear picture of the academic contributions and research activities of various institutions. Here, N = 281 (institutions); E = 221 (collaborative links); and D e n s i t y = 0.0056 (network density). The top 10 institutions working on digital watermarking, by publication volume, from 2004 to 2024 are listed in Table 2, which shows that the majority of top research institutes in digital watermarking are located in Taiwan, Singapore, and mainland China. Among them, institutions in mainland China have made particularly significant contributions, with the Chinese Acad Sci (180 papers), Nanjing Univ Informat Sci & Technol (102 papers), Liaoning Normal Univ (96 papers), and Hunan Univ (87 papers) demonstrating strong research output. In addition, Feng Chia Univ in Taiwan (153 papers) and Nanyang Technol Univ in Singapore (67 papers) have also established a solid research presence in this field. The collaboration network reveals that institutions with higher publication volumes tend to have more extensive collaborative ties and primarily engage in localized partnerships, leading to a strong geographic clustering effect. In contrast, institutions with lower publication volumes exhibit sparse collaboration or, in some cases, no collaboration at all. This indicates that research in this domain remains largely regionally focused, with a globalized academic collaboration network yet to fully develop. It is noteworthy that in the past five years, some emerging institutions have rapidly risen and had a certain influence on the field of digital watermarking. For example, the number of papers published in international journals by universities and research institutes such as Hangzhou Dianzi University, Shanghai University, and Jinan University has continued to climb since 2018, showing a strong latecomer advantage. This trend not only reflects the gradual expansion of the layout of digital watermarking technology research in China’s domestic universities, but also signals that the proliferation of research power in this field is to more emerging institutions. A deeper analysis suggests that China’s dominant position in this field is closely linked to its government policies, industry demands, and technological advancements. The Chinese government places a strong emphasis on data security, copyright protection, and geospatial information security, actively supporting research into digital watermarking. Additionally, industries such as cartography, satellite remote sensing, e-government, and short-video copyright protection have seen a rising demand for robust copyright protection measures. As a result, research institutions in mainland China have emerged as global leaders in digital watermarking research, driving scientific output and becoming an international influence in the field.

3.4. Country Cooperation Network Analysis

Figure 4 presents the regional collaboration network formed in the field of digital watermarking from 2004 to 2024, revealing the global research landscape. This study has identified 102 participating countries ( N = 102 ) and 118 collaborative links ( E = 118 ), with a network density of 0.0229. These findings indicate that although research on watermarking technology exhibits international dimensions, cross-country cooperation remains fragmented and has not yet resulted in a highly cohesive global research network. Further analysis reveals that certain transnational collaborations have been established between specific countries. For example, research institutions in China and Japan have cooperated closely on digital watermarking technology and have jointly published a number of high-impact publications [45,46,47]. A distinct geographical stratification can be seen in Table 3, which provides a list of the top 10 regions in digital watermark research, by publication volume, over the last 20 years. In recent years, the subject has been dominated by two new research powerhouses: PEOPLES R CHINA (3195 publications) and INDIA (1186 papers). Early technological pioneers, including South Korea (452 papers), the USA (857 papers), and Taiwan (922 papers), are still very active today, demonstrating their ongoing impact on the development of watermarking technologies. From a temporal perspective, the evolution of national contributions can be categorized into two distinct phases. The first wave (USA, TAIWAN, SOUTH KOREA, JAPAN) established a relatively mature research framework between 2004 and 2012, benefiting from early legislative initiatives such as the Digital Millennium Copyright Act (DMCA) enacted by the United States in 1998. This legislation spurred advancements in multimedia content protection, positioning these countries at the forefront of early watermarking research. The second wave (PEOPLES R CHINA, INDIA) has been the fastest-growing research force globally since 2013 and is largely driven by strong governmental support. Policies such as China’s 13th Five-Year National Informatization Plan (2016) and India’s Digital India Initiative (2015) have accelerated breakthroughs in multimedia security and geospatial data encryption, significantly expanding the research frontiers in these domains.

3.5. Keyword Co-Occurrence Clustering Analysis

A keyword co-occurrence analysis focuses on the co-occurrence of the keywords provided by the authors in the article and reveals the degree of association between and structural features of research topics by counting the frequency of their simultaneous appearance in the same article. High-frequency keywords are obtained by counting the total number of times each keyword is found in each article. CiteSpace can automatically extract and count the keywords, identify high-frequency keywords as network nodes according to the frequency of their occurrence, and draw a connecting line between the nodes with co-occurrence relationships, so as to construct a network of high-frequency keywords and form a keyword co-occurrence map, as shown in Figure 5. This map consists of 273 nodes ( N = 273 ) with 304 collaborative links ( E = 304 ), resulting in an overall network density of 0.082 ( D e n s i t y = 0.082 ), systematically revealing the research landscape of digital watermarking over the past two decades. Examples of keywords like “watermark”, “fragile watermarking”, and “authentication” highlight that early studies were on watermark extraction and embedding methods, their viability, and their applications in various transform domains. Research objectives changed toward improving robustness and optimizing watermarking algorithms as the need for multimedia security increased. Our co-occurrence analysis highlights “robust” and frequency-domain algorithms like “DCT” and “DWT” as central keywords of this phase, reflecting the expansion and integration of frequency-domain techniques into robust watermarking research. Scholars increasingly investigated the resilience of watermarks against compression, noise, and geometric attacks. Furthermore, advancements in artificial intelligence (AI) have introduced new dimensions to digital watermarking research. The focus has gradually transitioned from singular algorithmic improvements toward intelligent and multidimensional security frameworks. The emergence of keywords such as “deep learning”, “encryption”, “information security”, “steganography”, and “reversible data hiding” in recent years signifies that these areas have become crucial research frontiers in digital watermarking.
Cluster analyses can group nodes in a network based on their similarity to form several closely related sets. In CiteSpace, a cluster analysis can be performed on a keyword co-occurrence map to measure and categorize the relevance of keywords; help researchers identify subfields, theme groups, or concept clusters of a research field; and reveal the hotspots and frontiers of academic research. This study employs the log-likelihood ratio (LLR) algorithm for its keyword clustering analysis and visualizes the top 10 major clusters (Figure 6). To ensure the scientific rigor and reliability of the results, we utilized two clustering quality evaluation metrics: Q (Modularity) and S (Silhouette). The Q value assesses the modularity of the network, where Q > 0.3 indicates a significant clustering structure. The S value measures intra-cluster similarity, with S > 0.5 signifying that the clustering results are highly reliable. The analysis yields Q = 0.8456 and S = 0.9498 , confirming that the clustering structure is both reasonable and highly reliable. The following are the labels assigned to the top 10 keyword clusters: #0 deep learning, #1 audio watermarking, #2 reversible data hiding, #3 color image watermarking, #4 computational modeling, #5 quantum image processing, #6 copyright protection, #7 image authentication, #8 singular value decomposition, and #9 image encryption. These clusters fall into three main categories according to the clustering results generated by CiteSpace:
  • Applications in Copyright Protection and Data Integrity Verification (#6, #7): This research area took shape around 2005 and has remained steadily hot since then. With the exponential growth of digital content, integrating watermarking solutions into legal and technological frameworks has become a research hotspot. Utilizing frequency-domain methods to improve watermark tamper resistance is the main focus of recent research in this area.
  • Specific Application Domains of Watermarking Technology (#1, #2, #3, #9): This is an emerging area of research that has garnered increasing attention since 2010. Audio watermarking is widely employed in digital rights management, especially in the field of music, where the demand is significant. Reversible data hiding plays a crucial role in lossless information storage, particularly in medical imaging and remote sensing, as it has the technical advantage of lossless information restoration. Color image watermarking focuses on enhancing robustness in image-based applications. Image encryption primarily addresses data security and privacy protection.
  • Implementation Methods for Watermarking Technology (#0, #4, #5, #8): This is a cutting-edge direction that watermarking technology have evolved in in recent years. Deep learning is driving the advancement of intelligent watermarking algorithms, including watermark embedding and detection techniques based on convolutional neural networks (CNNs) and Generative Adversarial Networks (GANs). Other methodologies, such as computational modeling, quantum image processing, and singular-value decomposition, are also expanding the theoretical and technical foundations of watermarking research.
The analysis of the evolution of these trends further reveals that the early keyword clusters (e.g., #6, #7) are still active, but the focus on this research is relatively stable. However, clusters such as #0, #2, #9, etc., have significantly risen in importance in the past five years, indicating that the digital watermarking research driven by new technology is progressively evolving from application-oriented areas to intelligentization and securitization. On the other hand, some traditional areas such as #4, whose keyword activity has gradually decreased since 2015, show a declining trend.

3.6. Timeline Visualization of Keyword Co-Occurrence Clustering Analysis

The keyword clustering timeline is essentially an extension of the clustering structure in the time dimension, which is based on the keyword clustering results. The keywords in each cluster are sorted according to the time of their first appearance and arranged sequentially along the time axis. Along this axis, each timeline corresponds to a clustering unit, and the keywords labeled on the left side represent the core terms in that cluster, which are usually also the main nodes in the keyword co-occurrence network. Therefore, the timeline can be regarded as a combination of a clustering graph and a time zone graph, presenting both the clustering relationships among keywords and the dynamic process of the evolution of each topic over time. With the help of the timeline view, researchers are able to intuitively observe the development trajectory of different research topics and identify keywords that are active or prominent during a specific period of time, so as to reveal the process of the generation, continuation, and decline of research hotspots. This study uses a keyword clustering timeline to analyze pertinent academic trends and backgrounds in order to uncover the evolutionary trajectory of digital watermarking research (Figure 7). The CiteSpace timeline shows the evolution of the clustering results in the field of digital watermarking over time, providing insights into changing research hotspots, technological advancements, and their connections. The following phases can be used to roughly categorize the evolution of digital watermarking technology from a time standpoint:
  • Foundational Stage (2004–2016): During this phase, research primarily focused on the protection of image data. The representative cluster “#9 image encryption” emphasized chaotic systems and Fourier transform techniques, laying the groundwork for future watermarking advancements. Simultaneously, “#8 singular value decomposition” was widely applied and enhanced watermark robustness, gradually evolving into schemes suitable for medical imaging. Additionally, “#7 image authentication” research began to emerge, focusing on tamper detection and watermark recovery, with some studies incorporating neural network classification methods into their work. As the dissemination of digital content expanded, watermarking applications evolved beyond simple embedding and detection to include copyright protection and authentication. The “#6 copyright protection” cluster reflects the emergence of methods such as zero and semi-fragile watermarking, which began to be integrated into digital rights management systems, in the context of digital rights protection legislation and the rise of the content industry in the mid-to-late 2000s.
  • Multi-Type Watermarking Stage (2017–2020): This period witnessed significant advancements in reversible data hiding and multimedia watermarking. The “#3 color image watermarking” cluster expanded from grayscale to color images, progressing from algorithms and transform-based approaches to moment-based feature extraction (zernike moments, invariants), making watermarking more suitable for multimedia content protection. Meanwhile, “#2 reversible data hiding” emerged as a research hotspot, evolving from traditional watermarking techniques to lossless data hiding. Optimizations such as the discrete wavelet transform (DWT), predictions, and quality assessments were introduced, with recent studies extending to data mining, and their application has increased significantly in medical, military, and legal fields, which have strict requirements for data integrity. The focus of research in this phase has shifted to scenario-specific adaptations and utility optimization, and the driving factors behind this include the improvement of lossless standards for medical images and the development of the multimedia content industry, which requires a wide range of watermark technologies.
  • Intelligence-Driven Stage (2021–2024): In recent years, the rise of deep learning has accelerated the advancement of watermarking technology. The “#0 deep learning” cluster has become the core research direction in digital watermarking, evolving from early studies on the human visual system (HVS) to deep learning- and neural network-based techniques. The research trajectory of this cluster reflects a shift toward convolutional neural network (CNN)-driven privacy protection methods, signifying a transition from traditional security mechanisms to intelligent privacy-preserving technologies, driven by both technological and societal needs. This phase is also highly relevant to the implementation of data protection regulations such as the EU GDPR, prompting watermarking technology to be emphasized in privacy protection and data security compliance. In addition, the rapid development of the digital economy and digital assets has driven the demand for intelligent copyright protection solutions, providing both policy and industry support for research into this technology.

3.7. Keyword Burst Analysis

Burst words are created when there is an abrupt increase in keyword frequency at a particular time, highlighting the focus of researchers and new developments in technology. As shown in Figure 8, the development of digital watermarking technology can be divided into three phases through the visual analysis of emergent keywords, which reveal its expansion in different application areas. The predominance of keywords like digital watermarking, copyright protection, and video indicates that early research was mostly focused on image, video, and copyright protection. This period coincided with the rapid development of multimedia communication on the Internet and the growing problem of piracy and copyright infringement, which drove researchers to focus on how to embed robust watermarks using transform-domain methods (e.g., DCT, DWT) to efficiently identify and track the copyright attribution of digital content. Furthermore, the development of data hiding implies that researchers were actively looking for ways to improve security and invisibility by embedding strong but undetectable watermark signals in pictures and videos. It lays the foundation for the subsequent development of programs such as invisible watermarking and blind detection. These keyword bursts not only reveal the technical orientation of academic research at that time but also correspond to the industrialized need for digital media copyright protection, driven by policies and regulations (e.g., the Digital Millennium Copyright Act). The mid-term phase saw a steady shift in research priorities from basic watermarking techniques to improvements in robustness and capacity. During this time, segmentation, quantization, histogram modification, and the HVS were important areas of research. These patterns show a growing focus on watermarking strategies that strike a compromise between visual quality and imperceptibility, guaranteeing safe embedding and attack resistance. In addition, the prominence of image authentication shows that watermarking technology was gradually expanding to include data integrity verifications and tampering detection, enabling it to support financial credentials, legal evidence, and other scenarios that require high levels of originality in high their data. In the most recent phase, the rapid advancement of deep learning has become a key driving force in watermarking research. The burst analysis shows that deep learning exhibits the highest burst intensity (40.55), highlighting a growing research interest in integrating deep neural networks with watermarking algorithms to enhance performance. Simultaneously, the emergence of medical image as a burst term underscores the expanding use of watermarking technology in high-value data protection, particularly in the security and privacy preservation of medical images. Watermarking techniques in this domain serve as effective tools for preventing the unauthorized access and tampering of sensitive medical data. For example, by embedding reversible watermarks into medical images, not only can sensitive information be prevented from being leaked, but the integrity and traceability of the image’s content can also be ensured, meeting the dual requirements of medical compliance and privacy protection. These emerging applications further illustrate the versatility and potential of digital watermarking technology.

4. Discussion

This study analyzes 8621 relevant publications from 2004 to 2024, using CiteSpace to visualize research trends in digital watermarking. Through a descriptive analysis of publication timelines, key authors, and national contributions, it outlines the current state of the research in this field. Additionally, by employing a keyword co-occurrence analysis, clustering analysis, and temporal evolution mapping, this study identifies the research hotspots and development trajectory of digital watermarking.
As a crucial information security technique, digital watermarking has garnered academic attention since its early stages. With the proliferation of digital products, increasing demands for copyright protection, and advancements in policies, research in this domain has steadily grown, exhibiting an overall upward trend in publication volume. After 2017, the field entered a phase of rapid development, which peaked in 2023.
The global research landscape reveals a distinct geographical clustering effect in this field of research. Between 2004 and 2012, regions such as the USA, Taiwan, and South Korea laid the groundwork for digital watermarking research. However, since 2013, driven by policy incentives and industry demands, China and India have emerged as dominant research forces. Despite this progress, international collaborations remain relatively fragmented, and a highly interconnected global research network is lacking. At the author level, research has shifted from individual contributions to team collaborations, with high-impact researchers increasingly engaging in cooperative studies, reflecting a growing emphasis on collaborative research mechanisms.
In terms of research themes and content, the focus of digital watermarking research has evolved from its initial emphasis on copyright protection to a broader range of applications, including high-capacity information hiding and the integration of deep learning. More contemporary studies primarily explore copyright protection, data integrity verification, multimedia watermarking, and intelligent watermarking techniques. In particular, the introduction of deep learning has promoted the intelligent evolution of watermarking algorithms. In addition, the application of watermarking technology has been extended to medical imaging and other high-value data protection fields, showcasing its broad development prospects.

5. Conclusions

Over the past two decades, research in digital watermarking has transitioned from foundational theoretical development to multi-domain integration and is now advancing toward a stage that combines intelligence and high security with broad applicability. Early studies primarily focused on enhancing the robustness, imperceptibility, and computational efficiency of watermarking algorithms to improve their resilience against complex attacks. In recent years, driven by rapid advancements in information technology—particularly the emergence of deep learning and blockchain—digital watermarking has evolved toward a paradigm that prioritizes intelligence, efficiency, and security. For instance, integrating blockchain technology into digital watermarks provides a decentralized, tamper-proof mechanism for watermark verification, extending its applications beyond copyright protection to data provenance, distributed storage, and anti-tampering security. By leveraging smart contracts and blockchain ledgers, digital watermarking can serve as a digital credential, enabling end-to-end data authentication and traceability, ensuring the integrity and authenticity of data. This is particularly valuable in digital content transactions, copyright management, and electronic evidence preservation. Digital watermarking is expected to play an indispensable role in the future in important areas including privacy protection, digital copyright, and information security. It will also offer more dependable security for global digital development. However, while CiteSpace provides valuable insights into the evolution of scientific knowledge, it also has an inherent limitation in that it relies on co-citation data, which may lead to the neglect of emerging research topics that have not yet accumulated many citations. To mitigate this problem, we will continue to expand our data sources and extend the time span used in future studies, thereby enabling better time-series continuity and precision in our analysis. In addition, we will explore research in other languages, which will help us to conduct analyses with more temporal continuity and accuracy and build a more comprehensive knowledge graph.

Author Contributions

Conceptualization, M.W. (Mingkang Wu) and M.W. (Maofeng Weng); methodology, M.W. (Maofeng Weng) and E.M.; validation, W.Q., Y.D. and X.X.; formal analysis, X.X.; investigation, X.X.; resources, M.W. (Maofeng Weng); data curation, M.W (Mingkang Wu). and X.X.; writing—original draft preparation, M.W. (Mingkang Wu) and X.X.; writing—review and editing, W.Q., Y.D. and X.X.; visualization, M.W. (Maofeng Weng), E.M. and X.X.; funding acquisition, X.X. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (Grant NO.42101420).

Data Availability Statement

The data used in this study can be accessed from Web of Science (https://www.webofscience.com, accessed on 31 December 2024).

Acknowledgments

The authors gratefully acknowledge the anonymous reviewers for their insightful criticism, which definitely enhanced this work.

Conflicts of Interest

Authors Maofeng Weng, Wei Qu, and Eryong Ma were employed by the company “Northwest Engineering Corporation Limited”. The remaining authors declare that this research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

References

  1. Pham, G.N.; Ngo, S.T.; Bui, A.N.; Tran, D.V.; Lee, S.-H.; Kwon, K.-R. Vector map random encryption algorithm based on multi-scale simplification and Gaussian distribution. Appl. Sci. 2019, 9, 4889. [Google Scholar] [CrossRef]
  2. Ding, C.; Tang, J.; Deng, M.; Liu, H.; Mei, X. A local encryption method for large-scale vector maps based on spatial hierarchical index and 4D hyperchaotic system. Int. J. Geogr. Inf. Sci. 2024, 38, 2272–2300. [Google Scholar] [CrossRef]
  3. Ren, N.; Guo, S.; Zhu, C.; Hu, Y. A zero-watermarking scheme based on spatial topological relations for vector dataset. Expert Syst. Appl. 2023, 226, 120217. [Google Scholar] [CrossRef]
  4. Abubahia, A.; Cocea, M. Evaluating the topological quality of watermarked vector maps. Appl. Soft Comput. 2018, 71, 849–860. [Google Scholar] [CrossRef]
  5. Li, Y.; Zhang, L.; Wang, X.; Zhang, X.; Zhang, Q. A novel invariant based commutative encryption and watermarking algorithm for vector maps. ISPRS Int. J. Geo-Inf. 2021, 10, 718. [Google Scholar] [CrossRef]
  6. Wang, X.; Yan, H.; Zhang, L.; Zhang, X.; Li, P. An encryption algorithm for vector maps based on the Gaussian random and Haar transform. J. Spat. Sci. 2023, 68, 303–318. [Google Scholar] [CrossRef]
  7. Abubahia, A.; Cocea, M. Advancements in GIS map copyright protection schemes—A critical review. Multimed. Tools Appl. 2017, 76, 12205–12231. [Google Scholar] [CrossRef]
  8. Wang, X.; Yan, H.; Zhang, L. Vector map encryption algorithm based on double random position permutation strategy. ISPRS Int. J. Geo-Inf. 2021, 10, 311. [Google Scholar] [CrossRef]
  9. Qu, C.; Du, J.L.; Xi, X.; Tian, H.M.; Zhang, J. A hybrid domain-based watermarking for vector maps utilizing a complementary advantage of discrete fourier transform and singular value decomposition. Comput. Geosci. 2024, 183, 105515. [Google Scholar] [CrossRef]
  10. Li, Y.; Wei, D.; Zhang, L. Double-encrypted watermarking algorithm based on cosine transform and fractional Fourier transform in invariant wavelet domain. Inf. Sci. 2021, 551, 205–227. [Google Scholar] [CrossRef]
  11. Zhang, L.; Yan, H.; Zhu, R.; Du, P. Combinational spatial and frequency domains watermarking for 2D vector maps. Multimed. Tools Appl. 2020, 79, 31375–31387. [Google Scholar] [CrossRef]
  12. Cox, I.J.; Kilian, J.; Leighton, F.T.; Shamoon, T. Secure spread spectrum watermarking for multimedia. IEEE Trans. Image Process. 1997, 6, 1673–1687. [Google Scholar] [CrossRef] [PubMed]
  13. Xi, X.; Zhang, J.; Du, J.; Yang, Z. Desynchronization Attacks Resistant Watermarking for Remote Sensing Images Based on DWT-SVD and Normalized Feature Domain. Trans. GIS 2024, 28, 2705–2721. [Google Scholar] [CrossRef]
  14. Yan, H.; Zhang, L.; Yang, W. A normalization-based watermarking scheme for 2D vector map data. Earth Sci Inf. 2017, 10, 471–481. [Google Scholar] [CrossRef]
  15. Tang, Y.; Wang, S.; Wang, C.; Xiang, S.; Cheung, Y.-M. A highly robust reversible watermarking scheme using embedding optimization and rounded error compensation. IEEE Trans. Circuits Syst. Video Technol. 2023, 33, 1593–1609. [Google Scholar] [CrossRef]
  16. Wu, M.; Xi, X.; Du, J.; Zhang, X. Robust Reversible Watermarking Algorithm for Vector Maps Based on Virtual Triangle Feature Domain and Improved Quantization Index Modulation. Trans. GIS 2025, 29, e70036. [Google Scholar] [CrossRef]
  17. Wang, H.; Yao, H.; Qin, C.; Zhang, X. When Robust Reversible Watermarking Meets Cropping Attacks. IEEE Trans. Circuits Syst. Video Technol. 2024, 34, 13282–13296. [Google Scholar] [CrossRef]
  18. Hatoum, M.W.; Couchot, J.-F.; Couturier, R.; Darazi, R. Using deep learning for image watermarking attack. Signal Process. Image Commun. 2021, 90, 116019. [Google Scholar] [CrossRef]
  19. Pan, Z.; Xu, J.; Guo, Y.; Hu, Y.; Wang, G. Deep learning segmentation and classification for urban village using a worldview satellite image based on U-Net. Remote Sens. 2020, 12, 1574. [Google Scholar] [CrossRef]
  20. Berdik, D.; Otoum, S.; Schmidt, N.; Porter, D.; Jararweh, Y. A survey on blockchain for information systems management and security. Inf. Process. Manag. 2021, 58, 102397. [Google Scholar] [CrossRef]
  21. Jing, N.; Liu, Q.; Sugumaran, V. A blockchain-based code copyright management system. Inf. Process. Manag. 2021, 58, 102518. [Google Scholar] [CrossRef]
  22. Liang, W.; Zhang, D.; Lei, X.; Tang, M.; Li, K.-C.; Zomaya, A.Y. Circuit copyright blockchain: Blockchain-based homomorphic encryption for IP circuit protection. IEEE Trans. Emerg. Top. Comput. 2020, 9, 1410–1420. [Google Scholar] [CrossRef]
  23. Ren, N.; Wang, H.; Chen, Z.; Zhu, C.; Gu, J. A multilevel digital watermarking protocol for vector geographic data based on blockchain. J. Geovis. Spat. Anal. 2023, 7, 31. [Google Scholar] [CrossRef]
  24. Ren, N.; Zhao, Y.; Zhu, C.; Zhou, Q.; Xu, D. Copyright Protection Based on Zero Watermarking and Blockchain for Vector Maps. ISPRS Int. J. Geo-Inf. 2021, 10, 294. [Google Scholar] [CrossRef]
  25. Salman, T.; Zolanvari, M.; Erbad, A.; Jain, R.; Samaka, M. Security services using blockchains: A state of the art survey. IEEE Commun. Surv. Tutor. 2018, 21, 858–880. [Google Scholar] [CrossRef]
  26. Wang, B.; Jiawei, S.; Wang, W.; Zhao, P. Image copyright protection based on blockchain and zero-watermark. IEEE Trans. Netw. Sci. Eng. 2022, 9, 2188–2199. [Google Scholar] [CrossRef]
  27. Xu, Y.; Zhang, C.; Zeng, Q.; Wang, G.; Ren, J.; Zhang, Y. Blockchain-enabled accountability mechanism against information leakage in vertical industry services. IEEE Trans. Netw. Sci. Eng. 2020, 8, 1202–1213. [Google Scholar] [CrossRef]
  28. Zhu, P.; Hu, J.; Li, X.; Zhu, Q. Using blockchain technology to enhance the traceability of original achievements. IEEE Trans. Eng. Manag. 2021, 70, 1693–1707. [Google Scholar] [CrossRef]
  29. Kamaruddin, N.S.; Kamsin, A.; Por, L.Y.; Rahman, H. A review of text watermarking: Theory, methods, and applications. IEEE Access 2018, 6, 8011–8028. [Google Scholar] [CrossRef]
  30. Moulin, P. The role of information theory in watermarking and its application to image watermarking. Signal Process. 2001, 81, 1121–1139. [Google Scholar] [CrossRef]
  31. Tao, H.; Chongmin, L.; Zain, J.M.; Abdalla, A.N. Robust image watermarking theories and techniques: A review. J. Appl. Res. Technol. 2014, 12, 122–138. [Google Scholar] [CrossRef]
  32. De Vleeschouwer, C.; Delaigle, J.-F.; Macq, B. Invisibility and application functionalities in perceptual watermarking an overview. Proc. IEEE 2002, 90, 64–77. [Google Scholar] [CrossRef]
  33. Rashid, A. Digital watermarking applications and techniques: A brief review. Int. J. Comput. Appl. Technol. Res. 2016, 5, 147–150. [Google Scholar]
  34. Coatrieux, G.; Lecornu, L.; Sankur, B.; Roux, C. A review of image watermarking applications in healthcare. In Proceedings of the 2006 International Conference of the IEEE Engineering in Medicine and Biology Society, New York, NY, USA, 30 August–3 September 2006; pp. 4691–4694. [Google Scholar]
  35. Tsougenis, E.D.; Papakostas, G.A.; Koulouriotis, D.E.; Tourassis, V.D. Performance evaluation of moment-based watermarking methods: A review. J. Syst. Softw. 2012, 85, 1864–1884. [Google Scholar] [CrossRef]
  36. Peng, F.; Lei, Y.-Z.; Long, M.; Sun, X.-M. A reversible watermarking scheme for two-dimensional CAD engineering graphics based on improved difference expansion. Comput Aided Des. 2011, 43, 1018–1024. [Google Scholar] [CrossRef]
  37. Boenisch, F. A systematic review on model watermarking for neural networks. Front. Big Data 2021, 4, 729663. [Google Scholar] [CrossRef]
  38. Chou, C.-M.; Tseng, D.-C. Technologies for 3d model watermarking: A survey. Int. J. Comput. Sci. Netw. Secur. 2007, 7, 328–334. [Google Scholar]
  39. Chen, C. CiteSpace II: Detecting and visualizing emerging trends and transient patterns in scientific literature. J. Am. Soc. Inf. Sci. Technol. 2006, 57, 359–377. [Google Scholar] [CrossRef]
  40. Sun, Y.; Wu, S.; Gong, G. Trends of research on polycyclic aromatic hydrocarbons in food: A 20-year perspective from 1997 to 2017. Trends Food Sci. Technol. 2019, 83, 86–98. [Google Scholar] [CrossRef]
  41. Gao, J.; Li, J.; Geng, Y.; Yan, Y. Research progress of health education for adolescents based on CiteSpace analysis. Environ. Dev. Sustain. 2024, 1–40. [Google Scholar] [CrossRef]
  42. Li, Y.; Li, M.; Sang, P. A bibliometric review of studies on construction and demolition waste management by using CiteSpace. Energy Build. 2022, 258, 111822. [Google Scholar] [CrossRef]
  43. Zhang, K.; Liu, F.; Zhang, H.; Duan, Y.; Luo, J.; Sun, X.; Wang, M.; Ye, D.; Wang, M.; Zhu, Z. Trends in phytoremediation of heavy metals-contaminated soils: A web of science and CiteSpace bibliometric analysis. Chemosphere 2024, 352, 141293. [Google Scholar] [CrossRef] [PubMed]
  44. Zhong, D.; Li, Y.; Huang, Y.; Hong, X.; Li, J.; Jin, R. Molecular mechanisms of exercise on cancer: A bibliometrics study and visualization analysis via CiteSpace. Front. Mol. Biosci. 2022, 8, 797902. [Google Scholar] [CrossRef] [PubMed]
  45. Meng, Z.; Morizumi, T.; Miyata, S.; Kinoshita, H. Design scheme of copyright management system based on digital watermarking and blockchain. In Proceedings of the 2018 IEEE 42nd Annual Computer Software and Applications Conference (COMPSAC), Tokyo, Japan, 23–27 July 2018; pp. 359–364. [Google Scholar]
  46. Lu, W.; Chen, Z.; Li, L.; Cao, X.; Wei, J.; Xiong, N.; Li, J.; Dang, J. Watermarking based on compressive sensing for digital speech detection and recovery. Sensors 2018, 18, 2390. [Google Scholar] [CrossRef]
  47. Tseng, K.-K.; Na, Q.; Lin, R.F.-Y. A Document Management System with Multi-biometric Watermarking for Educational Purpose. In Proceedings of the International Conference on Intelligent Information Hiding and Multimedia Signal Processing, Kitakyushu, Japan, 16–18 December 2022; pp. 265–275. [Google Scholar]
Symmetry 17 00871 g001
Figure 1. Number of publications generated by year.
Figure 1. Number of publications generated by year.
Symmetry 17 00871 g001
Symmetry 17 00871 g002
Figure 2. Author collaboration network.
Figure 2. Author collaboration network.
Symmetry 17 00871 g002
Symmetry 17 00871 g003
Figure 3. Institution collaboration network.
Figure 3. Institution collaboration network.
Symmetry 17 00871 g003
Symmetry 17 00871 g004
Figure 4. Region collaboration network.
Figure 4. Region collaboration network.
Symmetry 17 00871 g004
Symmetry 17 00871 g005
Figure 5. Network of co-occurring keywords.
Figure 5. Network of co-occurring keywords.
Symmetry 17 00871 g005
Symmetry 17 00871 g006
Figure 6. Network of co-occurring keyword clusters.
Figure 6. Network of co-occurring keyword clusters.
Symmetry 17 00871 g006
Symmetry 17 00871 g007
Figure 7. Timeline of keyword co-occurrences.
Figure 7. Timeline of keyword co-occurrences.
Symmetry 17 00871 g007
Symmetry 17 00871 g008
Figure 8. Top keywords with the strongest citation bursts. (Light blue: Not yet appeared; Dark blue: Appeared; Red: Burst period).
Figure 8. Top keywords with the strongest citation bursts. (Light blue: Not yet appeared; Dark blue: Appeared; Red: Burst period).
Symmetry 17 00871 g008
Table 1. Top 10 authors with the highest number of publications.
Table 1. Top 10 authors with the highest number of publications.
Total PublicationsAuthorsTotal PublicationsAuthors
108Chang, Chin-Chen27Li, Li
60Zhang, Xinpeng24Zhang, Weiming
48Singh, Amit Kumar23Zhou, Ri-Gui
39Su, Qingtang23Ma, Bin
29Wang, Chunpeng23Li, Jingbing
Table 2. Top 10 institsutions with the highest number of publications.
Table 2. Top 10 institsutions with the highest number of publications.
Total PublicationInstitutionsTotal PublicationsInstitutions
180Chinese Acad Sci86Harbin Inst Technol
153Feng Chia Univ80Shanghai Univ
102Nanjing Univ Informat Sci & Technol71Shanghai Jiao Tong Univ
96Liaoning Normal Univ69Sun Yat Sen Univ
87Hunan Univ67Nanyang Technol Univ
Table 3. Top 10 regions with the highest number of publications.
Table 3. Top 10 regions with the highest number of publications.
Total PublicationsCountries/RegionsTotal PublicationsCountries/Regions
3195PEOPLES R CHINA294JAPAN
1186INDIA277SAUDI ARABIA
922TAIWAN266IRAN
857USA265FRANCE
452SOUTH KOREA246ENGLAND
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Weng, M.; Qu, W.; Ma, E.; Wu, M.; Dong, Y.; Xi, X. Bibliometric Analysis of Digital Watermarking Based on CiteSpace. Symmetry 2025, 17, 871. https://doi.org/10.3390/sym17060871

AMA Style

Weng M, Qu W, Ma E, Wu M, Dong Y, Xi X. Bibliometric Analysis of Digital Watermarking Based on CiteSpace. Symmetry. 2025; 17(6):871. https://doi.org/10.3390/sym17060871

Chicago/Turabian Style

Weng, Maofeng, Wei Qu, Eryong Ma, Mingkang Wu, Yuxin Dong, and Xu Xi. 2025. "Bibliometric Analysis of Digital Watermarking Based on CiteSpace" Symmetry 17, no. 6: 871. https://doi.org/10.3390/sym17060871

APA Style

Weng, M., Qu, W., Ma, E., Wu, M., Dong, Y., & Xi, X. (2025). Bibliometric Analysis of Digital Watermarking Based on CiteSpace. Symmetry, 17(6), 871. https://doi.org/10.3390/sym17060871

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop