Year: 2025
Type: article
Abstract: With the advancement of technology and the growth of human demand, pedestrian re-identification is a key technology of intelligent systems and plays an important role in daily life. Traditional vision methods have certain limitations, but the millimeter-wave-vision fusion system, which complements the advantages of cameras and millimeter-wave radars, plays a greater role and is attracting widespread attention. This paper first introduces the effects of vision camera and millimeter-wave radar camera on human re-identification in a single mode, and then discusses the detailed processing required to fuse the data of the two modes, including the key technologies of sensor settings, data synchronization and sensor calibration. We also review the classification and evolution of millimeter-wave radar and visual fusion pedestrian re-identification methods, including data-level, feature-level and decision-level methods, and review the previous research methods, which will provide great inspiration for future research. Finally, this paper discusses the typical applications of millimeter-wave vision fusion systems, as well as the key technologies and potential challenges of fusing millimeter-wave radar and visual data, and looks forward to future research directions.
Authors: Qingyuan Yang, Zhipeng Quan, Jingxuan Li, Tingyv Jiang, Zhihao Deng, Xinran Qiu, Zhengjie Wang
Institution: Shandong University of Science and Technology
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4411458877