基于手机传感器的计步方法研究与系统实现

发布时间：2018-08-07 19:58

【摘要】：随着都市生活、工作节奏的不断加快以及人们饮食习惯的改变,都市健康问题不断加剧。实时检测步数可以帮助人们实时掌握锻炼情况,保持适度的日常运动,以增强身体素质,进而达高免疫力、预防慢性疾病和职业病的目的。现有的计步方法虽然表现出不错的准确率,但是对不规则步伐的识别度不高,并且抗干扰性也不够突出,一些非行走动作会对计步检测结果造成明显干扰并造成计步结果虚高。本文根据人们对日常运动检测和计步抗干扰的需求,研究并设计了一种具有良好抗干扰性的计步检测方法,包括对行走特征的选取、阈值的确定与判定流程设计以及计步系统实现。具体的研究工作如下:(1)通过分析人体行走时加速度的变化状况,选取了与行走动作最为相关的三个特征:幅度特征,以表示运动时的力度变化范围;角度特征,以反映人体行走运动式方向的变化情况;距离特征,以表示行走运动时人体的位移变化。进一步对三个特征,分别设计了特征计算方法。(2)幅度、角度、距离三个特征的确定对提高计步方法的抗干扰性至关重要。首先对三个特征,通过对比干扰动作与行走动作时域特点,建立了初始阈值设定原理,然后通过实验分别确定了三个特征的初始阈值并设计了阈值校正方法。(3)结合对干扰动作过滤方法的研究设计了一种计步方法,通过Android平台实现了计步系统,并把本计步器和使用其他计步方法的计步产品进行了对比测试。实验结果显示,本计步系统在正常行走的情形下具有合格的识别效率,对不连续步伐识别的准确率提高了10%-18%,对原地踏步、晃动手机的非行走行为干扰识别的误差率缩小到6%~12%,体现出了良好的抗干扰性。
[Abstract]:With the rapid development of urban life, the pace of work and the change of people's eating habits, urban health problems are becoming more and more serious. The real-time detection of steps can help people to grasp the situation of exercise in real time and maintain moderate daily exercise in order to enhance their physical quality and achieve the purpose of high immunity and prevention of chronic and occupational diseases. Although the existing methods show good accuracy, but the recognition of irregular steps is not high, and the anti-jamming is not enough, some non-walking actions will cause obvious interference to the step detection results and result in the false height of the steps. In this paper, according to people's demand for daily motion detection and anti-jamming, a method of step detection with good anti-interference is studied and designed, including the selection of walking features. The determination of threshold and the design of decision flow, and the realization of step-counting system. The specific research work is as follows: (1) by analyzing the change of acceleration in human walking, we select three characteristics that are most relevant to walking motion: amplitude feature to indicate the range of dynamics change while moving, angle characteristic, In order to reflect the movement direction of the human body changes, distance characteristics, to represent the movement of the human body displacement changes. Furthermore, the feature calculation method is designed for the three features. (2) the determination of the amplitude, angle and distance is very important to improve the anti-interference ability of the step calculation method. First of all, the principle of initial threshold setting is established by comparing the time domain characteristics of interference action with that of walking action. Then the initial threshold of the three characteristics is determined by experiments and the threshold correction method is designed. (3) combined with the research of the interference action filtering method, a step counting method is designed, and the step counting system is realized through the Android platform. The pedometer and other step-counting products are compared and tested. The experimental results show that the system has a qualified recognition efficiency under the condition of normal walking, and the accuracy of the recognition of discontinuous steps is improved by 10 to 18, and the standing step is fixed. The error rate of non-walking interference recognition of shaking mobile phone is reduced to 6 / 12, which shows good anti-interference.
【学位授予单位】：江苏大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TP212

【参考文献】