Research status and application prospects of biometric identification from unimodal to multimodal perspectives
- VernacularTitle:单模态至多模态生物特征身份识别研究现状与应用展望
- Author:
Jiahui CHEN
1
;
Shuhui GAO
1
;
Hongmin YUAN
1
;
Guirong WANG
1
Author Information
- Publication Type:Journal Article
- Keywords: biometric identification; forensic science; unimodal identification; multimodal identification; neural network; biometric fusion
- From: Journal of Xi'an Jiaotong University(Medical Sciences) 2025;46(4):551-561
- CountryChina
- Language:Chinese
- Abstract: Multimodal biometric identification technology,which combines multiple biometric methods,is being increasingly used in forensic science due to its advantages such as difficulty of forgery and great accuracy.However,the current identification mainly relies on unimodal biometrics including facial recognition,hand recognition,iris recognition,gait recognition,and voice recognition.Thus,to a large extent,the accuracy of identification depends on the quality of the unimodal data,which faces multiple challenges.In contrast,multimodal biometrics has more obvious advantages in the field of identification,which can not only effectively resist attacks,but also enrich the feature representation with complementary information from multimodal sources,mitigate the impact of environment on the identification performance,and enhance the robustness of the system.To this end,this paper combs through the relevant work in this field,and comprehensively reviews the current research status and development trend of biometric identification technology.First,based on bibliometrics,the paper combs through the relevant research results in the past ten years,analyzes,summarizes,and discusses the commonly used deep learning models in this field,respectively,from unimodal recognition to multi-modal fusion recognition.Then the paper discusses the future development direction of the multimodal biometric identification field in the light of the practical needs of court science and provides reference for identification research.
