基于FCM聚类和卷积神经网络的跌倒识别算法

doi:10.16337/j.1004-9037.2021.04.013

首页 > 按月查看>2021年第4月 >746-755. DOI:10.16337/j.1004-9037.2021.04.013

基于FCM聚类和卷积神经网络的跌倒识别算法
DOI:
                        10.16337/j.1004-9037.2021.04.013
                    
作者:
                        
                        
                    
作者单位:泰州职业技术学院机电技术学院，泰州 225300
作者简介:
通讯作者:
基金项目:江苏省高校自然科学研究面上(20KJD510008)资助项目；泰州市科技支撑计划(TS201817)资助项目。

Fall Recognition Algorithm Based on FCM Clustering and Convolutional Neural Network

Author:

Affiliation:

College of Electromechanical Technology, Taizhou Polytechnical College, Taizhou 225300,China

Fund Project:

摘要

图/表

访问统计

参考文献

相似文献

引证文献

资源附件

摘要:

为了提高传统跌倒检测系统的识别准确度和运算速度，减小误报率和漏报率，本文提出了一种基于模糊C均值（Fuzzy C-means， FCM）聚类算法和卷积神经网络算法的实时跌倒检测算法。该算法以深度视觉传感器为数据获取源，提取聚类中心点速度、高度、加速度以及夹角为跌倒识别特征向量，采用阈值分析和机器算法相结合的方式实现人体跌倒识别。实验表明，该算法的识别精度达到99%，运算速度为0.178 s，相对于传统算法具有更高的识别精度和运算速度。

Abstract:

In order to improve the recognition accuracy and operation speed of the traditional fall detection system and reduce the false alarm rate and the missing alarm rate， a real-time fall detection algorithm based on fuzzy C-means （FCM） clustering algorithm and convolutional neural network algorithm is proposed. The algorithm takes the depth vision sensor as the data acquisition source， extracts the velocity， the height， the acceleration， and the angle of the cluster center point as the fall recognition feature vector， and uses the combination of threshold analysis and machine algorithm to realize human fall recognition. The experimental results show that the recognition accuracy of the algorithm reaches 99% and the operation speed is 0.178 s， which is higher than those of the traditional algorithm.

表 1 各类算法模型在相同样本数据下的检测结果对比Table 1 Comparison of detection results of various algorithm models under the same sample data

表 2 Table 2 Performance of each scheduling strategy (Scenario 1)

图1 Flowchart of EGAFig.1

图2 Chromosome codingFig.2

图3 Two-point crossoverFig.3

图4 Trend of upper / lower level programming model objective function value in simultaneous iteration process (Scenario 1)Fig.4

图5 Trend of upper / lower level programming model objective function value in simultaneous iteration process (Scenario 2)Fig.5

图6 Optimal departure sequences and FCFS departure sequences for departure flights (Scenario 2)Fig.6

图7 Trend of upper / lower level programming model objective function value in simultaneous iteration process (Scenario 3)Fig.7

图8 Trend of upper / lower level programming model objective function value in simultaneous iteration process (Scenario 4)Fig.8

图1 系统结构Fig.1 System structure

图2 人体25个骨骼关节点分布图Fig.2 Distribution of 25 joints in human body

图3 人体站立时改进FCM和传统FCM算法比较Fig.3 Comparison of improved FCM and traditional FCM algorithms for a standing state

图4 人体蹲下时改进FCM和传统FCM算法比较Fig.4 Comparison between improved FCM and traditional FCM algorithms for a squatting state

图5 不同人体行为下骨架节点聚类中心结果显示Fig.5 Results of skeleton node cluster center for different human behaviors

图7 不同动作下两聚类中心速度变化曲线Fig.7 Velocity variation of two cluster centers for different actions

图8 三个中心点几何关系图Fig.8 Geometric relationship of three center points

图10 CNN跌倒检测模型Fig.10 Fall detection model based on CNN

图11 CNN训练步长和准确率之间的关系Fig.11 Relationship between CNN training step and accuracy

图12 跌倒检测流程Fig.12 Fall detection process

表 3 Table 3 Performance of each scheduling strategy (Scenario 3)

参考文献

相似文献

引证文献

引用本文

朱艳,李曙生,谢忠志.基于FCM聚类和卷积神经网络的跌倒识别算法[J].数据采集与处理,2021,36(4):746-755

复制

文章指标

点击次数:
下载次数:

历史

收稿日期:2020-09-09
最后修改日期:2021-03-05
录用日期:
在线发布日期: 2021-07-25

引用本文

分享

文章指标

历史