基于柔性太阳能电池和超薄水凝胶薄膜的手势识别  

Gesture recognition based on flexible solar cells and ultrathin hydrogel film

作　　者：吴如成 丁文伯 徐晓敏 宋林琦 徐伟涛 WU Rucheng;DING Wenbo;XU Xiaomin;SONG Linqi;XU Weitao(Shenzhen Research Institute,City University of Hong Kong,Shenzhen 518057,China;Department of Computer Science,City University of Hong Kong,Hong Kong 999077,China;Tsinghua Shenzhen International Graduate School,Shenzhen 518071,China)

机构地区：[1]香港城市大学深圳研究院,广东深圳518057 [2]香港城市大学计算机学院,中国香港999077 [3]清华大学深圳国际研究生院,广东深圳518071

出　　处：《电信科学》2023年第7期109-115,共7页Telecommunications Science

基　　金：国家自然科学基金资助项目(No.62101471);香港城市大学深圳研究院和中国香港特别行政区研究资助局资助项目(No.CityU21201420,No.City U11201422)

摘　　要：人们在日常生活中会用到各种各样的手势,如何将不同的手势与现有的智能可穿戴设备等结合起来对生活质量的提升有不可或缺的作用,利用太阳能相关设备的光电转换特性可以很好地解决手势识别和设备能耗问题。在柔性太阳能电池与手势识别结合的研究中,采集了5种常用手势数据,进行了Z-Score、低通滤波、滑动窗口等信号处理方法,并利用随机森林、支持向量机和神经网络等对其进行分类,成功地在小样本的基础上实现了100%的预测精度,可以证明该方法在手势识别的应用中有显著优势。Daily life involves various gestures,and combining these with smart wearable devices is crucial for im-proving quality of life.One effective solution to the challenges of gesture recognition and device energy consumption isutilizing the photoelectric conversion characteristics of solar energy-related devices.The data of five commonly used gestures were collected in the research of the combination of flexible solar cells and gesture recognition.Z-Score,low-pass filter,sliding window techniques for signal processing were applied,and successfully achieved 100%predicted accuracy using random forest,support vector machine and neural network algorithms even with small samples which showed that this method had significant advantages in the application of gesture recognition.

关 键 词：柔性太阳能电池 超薄水凝胶薄膜 手势识别 机器学习 智能可穿戴设备 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]