多模态下的手写识别信息特征精确预测研究  

Research on accurate prediction of handwriting recognition information features in multi-modality

作　　者：曹意 李建辉 刘辉 陈瀛 吴倩倩 CAO Yi;LI Jianhui;LIU Hui;CHEN Ying;WU Qianqian(Foshan Polytechnic,Foshan 528137,China;Guangdong University of Foreign Studies,Guangzhou 510000,China)

机构地区：[1]佛山职业技术学院,广东佛山528137 [2]广东外语外贸大学,广东广州510000

出　　处：《无线互联科技》2025年第1期88-93,共6页Wireless Internet Science and Technology

基　　金：广东省普通高校工程技术研究(开发)中心,项目编号:2024GCZX028。

摘　　要：近年来,随着人工智能、可穿戴和移动设备的快速发展,空中写作逐渐成为新型人机交互(Human-Computer Interaction,HCI)的研究热点。最新的研究均阐述了基于Wi-Fi信道状态信息(Channel State Information,CSI)的空中手写识别技术。该技术具有灵活部署和更少设备成本等硬性约束,在智能家居、医疗保健等领域有着广阔的应用前景。但是当测试环境发生变化,该技术在Wi-Fi环境下所能达到的识别性能迅速下降,此时,单纯使用以CSI为基础的手写识别技术存在较大的局限性。为此,文章引入一种与环境无关的手环加速度模态,通过对CSI模态和手环加速度模态进行融合,优化了信号处理方法,从而实现空中手写信息有效特征的精准预测,解决了环境变化带来的手写识别研究的局限性问题。In recent years,with the rapid development of artificial intelligence,wearable and mobile devices,in-flight writing has gradually become a research hotspot of new human-computer interaction(HCI).The latest researches all expound the in-air handwriting recognition technology based on Wi-Fi Channel State Information(CSI),which has hard constraints such as flexible deployment and less equipment cost,and has a wide application prospect in smart home,healthcare and other fields.However,when the test environment changes,the recognition performance that can be achieved in Wi-Fi environment decreases rapidly.At this time,there are great limitations in the research of handwriting recognition based on CSI alone.Therefore,this paper introduced an environment-independent wristband acceleration mode.Through the fusion of CSI mode and wristband acceleration mode,the signal processing method was optimized,so as to realize the accurate prediction of effective features of aerial handwriting information and solve the limitations of handwriting recognition research caused by environmental changes.

关 键 词：人机交互 空中写作 Wi-Fi CSI信息 加速度信息 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]