基于可穿戴式设备的智能家电系统的研究与实现

发布时间：2018-05-31 20:04

本文选题：可穿戴式设备 + 手势识别　；参考：《中国海洋大学》2015年硕士论文

【摘要】：近年来,随着人们生活品质的提高,智能家电系统的控制与研究倍受关注,成为最活跃的研究方向之一。而手势是人类日常生活中必不可少的一部分,是人与人之间一种广泛的交流形式,在人机交互中,它逐步成为新兴的交互方式。基于现有的理论和研究,本文经过分析和总结,实现了一种基于隐马尔可夫模型的手势识别,通过对手势的识别达到控制家用电器的目的。本文以手势识别作为研究对象,对相关理论方法展开了系统的分析,从用户的实际需求出发,设计完成了基于可穿戴式设备的智能家电系统。本文主要从以下几个方面展开工作：(1)特征提取。特征提取主要包括时间域特征和频率域特征。由于时间域特征比较直观,而且容易提取,其性价比比提取频率域特征要高,因此本文选择提取时间域特征。最终提取出的时间域特征包括：均值、合成加速度、方差、振幅、加速度最大轴、波峰数、峰值距离、均方根特征和信号幅度区域。(2)基于隐马尔可夫模型的手势识别过程。本文首先通过实验采集提取特征数据,将特征数据分为六大组,包括上、下、左、右、前、后,然后通过HMM学习得出隐马尔可夫模型。拥有隐马尔可夫模型后,使用序列后向特征选择方法进行后项选择特征,选择出实验需要的特征,并建立出完善的隐马尔可夫模型。最终本文选择出的特征有五个,分别是：均值,振幅、加速度最大轴、波峰数以及均方根特征。最后,用预留出的测试数据对HMM模型进行测试,得到结果。以后使用此隐马尔可夫模型即可识别手势。本文通过进行实验模拟,验证了算法的有效性,对实现智能家电系统的控制提供了强有力的证据,可以为今后智能家电行业的发展提供一定的理论支持,使得人与电器交互更加的人性化、智能化、舒适化,让家电更加地认识用户、懂得用户的需求、了解用户的肢体语言。
[Abstract]:In recent years, with the improvement of people's quality of life, the control and research of intelligent home appliance system have attracted much attention, and become one of the most active research directions. Gesture is an indispensable part of human daily life, is a kind of extensive communication between people, in human-computer interaction, it has gradually become a new way of interaction. Based on the existing theory and research, this paper analyzes and summarizes a kind of gesture recognition based on hidden Markov model, through which the purpose of controlling household appliances is achieved. In this paper, gesture recognition is taken as the research object, the related theories and methods are systematically analyzed, and the intelligent appliance system based on wearable devices is designed and completed according to the actual needs of users. This paper mainly works on the following aspects: 1) feature extraction. Feature extraction mainly includes time domain feature and frequency domain feature. Because the feature of time domain is more intuitive and easy to extract, and its ratio of performance to price is higher than that of frequency domain, this paper chooses to extract the feature of time domain. The extracted time domain features include: mean, synthetic acceleration, variance, amplitude, maximum axis of acceleration, peak number, peak distance, root mean square feature and signal amplitude region. 2) gesture recognition process based on hidden Markov model. In this paper, the feature data are collected and extracted through experiments, and the feature data are divided into six groups: top, bottom, left, right, front and back. Then the hidden Markov model is obtained by HMM learning. After possessing the hidden Markov model, the sequential backward feature selection method is used to select the features needed in the experiment, and a perfect hidden Markov model is established. Finally, there are five features selected in this paper: mean, amplitude, maximum axis of acceleration, number of peaks and root mean square (RMS). Finally, the HMM model is tested with the reserved test data and the results are obtained. This hidden Markov model can be used to recognize gestures. Through the experimental simulation, the validity of the algorithm is verified, which provides a strong evidence for the realization of intelligent home appliance system control, and can provide certain theoretical support for the development of intelligent home appliance industry in the future. It makes the interaction between people and electrical appliances more humanized, intelligent and comfortable. It makes household appliances more aware of users, understand their needs, and understand their body language.
【学位授予单位】：中国海洋大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TM925.0

【相似文献】