摘要
针对人体行为事件,研究了多传感器数据采集模型和扩展卡尔曼滤波优化算法的应用。系统将加速度传感器分量数据映射为三维加速度空间,并与压力传感器数据结合建立四维实时数据采集空间。基于系统的模型特征,提出了非线性系统下的扩展卡尔曼滤波算法。系统利用优化算法对数据实现最优估计,并依据传感器信噪比对优化数据进一步修正,之后在系统设定的传感器信任级别和融合权重的基础上完成人体行为识别。实验结果表明,本文算法可以提高数据空间的精度和平滑度,可对人体行为进行有效识别。
By studying the model of multi-sensor data acquisition and Kalman filter, a new nonlinear algorithm for human action recognition is proposed in this paper. The acceleration sensor data is mapping to a three-dimensional space, which combined with pressure sensor data to form the four-dimensional data space. Then, the extended Kalman filter is used to process the combined data and revise the processed data. After that, system identifies human action based on trust level and weight fusion. The simulation results demonstrated that the algorithm can improve the accuracy of the sensor and effective identify the human action.
出处
《电子设计工程》
2016年第2期15-17,24,共4页
Electronic Design Engineering
基金
广东省教育厅科技创新项目(2013KJCX0178)
东莞市科技计划项目基金(2012108102007)