A survey of handwriting synthesis from 2019 to 2024: A comprehensive review

被引:0
作者
Diaz, Moises [1 ]
Mendoza-Garcia, Andrea [1 ]
Ferrer, Miguel A. [1 ]
Sabourin, Robert [2 ]
机构
[1] Univ Palmas Gran Canaria, Inst Univ Desarrollo Tecnol & Innovac Comunicac, Campus Tafira, Las Palmas Gran Canaria, Spain
[2] Univ Quebec, Ecole Technol Super, Montreal, PQ, Canada
关键词
Handwriting; Synthesis; Generation; Online; Offline; Review; SIGNATURE; REPRESENTATION; GENERATION; MODEL;
D O I
10.1016/j.patcog.2025.111357
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Handwriting, as a uniquely human skill, contributes to fine motor development and cognitive growth. Beyond mere functionality, handwriting carries individuality and subtle emotional nuances, evoking feelings of intimacy and authenticity. Consequently, the generation of synthetic handwritten manuscripts should not only prioritize the production of legible text, but also seek to enhance personalization and authenticity in digital communication. This enhancement renders handwriting synthesis invaluable in domains such as digital marketing and e-learning. Notably, handwriting synthesis plays a pivotal role in forensic science, particularly in signature verification, to bolster security and prevent fraud. Additionally, it has the potential to enhance accessibility, particularly for individuals with disabilities, and assist in health monitoring among elderly populations. Motivated by the significance of handwriting synthesis, this paper conducts a comprehensive literature review on the synthetic generation of handwriting and signatures. By examining research from 2019 to 2024, we categorize methods of synthesis, evaluate synthetic handwriting quality, and explore practical applications. Furthermore, we provide insights into publicly available code resources and emerging synthetic databases.
引用
收藏
页数:17
相关论文
共 132 条
  • [1] H.R. 8152, (2022)
  • [2] Election U.S., (2023)
  • [3] Abdollahzadeh M., Et al., A survey on generative modeling with limited data, few shots, and zero shot, (2023)
  • [4] Al-Suhaibani D., Al-Shargabi A.A., Cluster GAN-based model for signature generation and verification, 2023 3rd Int. Conf. on Computing and Information Technology, ICCIT, pp. 249-254, (2023)
  • [5] Alonso E., Moysset B., Messina R., Adversarial generation of handwritten text images conditioned on sequences, 2019 Int. Conf. on Document Analysis and Recognition, ICDAR, pp. 481-486, (2019)
  • [6] Arab N., Nemmour H., Chibani Y., A new synthetic feature generation scheme based on artificial immune systems for robust offline signature verification, Expert Syst. Appl., 213, (2023)
  • [7] Archibald T., Poggemann M., Chan A., Martinez T., Trace: a differentiable approach to line-level stroke recovery for offline handwritten text, 16th Int. Conf. on Document Analysis and Recognition, ICDAR, pp. 414-429, (2021)
  • [8] pp. 7564-7573, (2018)
  • [9] Barrere K., Soullard Y., Lemaitre A., Couasnon B., Training transformer architectures on few annotated data: an application to historical handwritten text recognition, Int. J. Doc. Anal. Recognit. (IJDAR), pp. 1-14, (2024)
  • [10] Bauer A., Et al., Comprehensive exploration of synthetic data generation: A survey, (2024)