改进随机森林及其在人体姿态识别中的应用

针对随机森林算法静态性、容易陷入局部最优等问题，提出了一种蜜蜂交配优化的随机森林算法，并将该算法应用于基于加速度传感器的人体姿态识别。设计了一套以三轴加速度传感器MMA7260与无线通信模块CC2430相结合的数据采集系统，采集了五种日常行为和一种异常行为；从加速度值中提取了近斜率、前后差、均值、均方根和信号幅值面积5类特征矢量；采用改进的随机森林算法训练行为模型和进行分类识别。实验结果表明：该算法能有效地识别六种行为，具有较高的分类预测准确率和行为识别率，且具有较强的稳定性、鲁棒性、全局寻优和抗噪声能力。

Improved random forest algorithm and its application in human posture patterns recognition

Adapting to the problem on the static property and easy convergence to local optima of traditional random forest algorithm, a honey-bee mating optimization random forest algorithm is proposed, and it is used to recognize the human motion patterns that is based on an acceleration sensor. A data collect system that is consist of a 3D acceleration sensor MMA7260 and wireless communication module CC2430 is designed, and is used to collect five daily activity and one abnormal behavior. Then, five features（Approximation Slope, Front and Rear Subtract, mean, RMS and SMA） are extracted from the acceleration signal. Finally, the improved random forest algorithm is adopted as a classifier. The experimental results have confirmed that the proposed algorithm is effective to recognize the six activity, meanwhile, the algorithm has achieved high classification prediction accuracy and recognition rate, and it has stronger stabile, robustness, ability of global optimal and anti-noise.