Abstract: To improve the smart phone built-in acceleration sensor personnel recognition rate, a method of combinatorial classifier recognition based on information segmentation is proposed. Firstly, according to the characteristics of pedestrian walking acceleration, a division method based on HMM（Hidden Markov Model） is proposed. The pedestrian acceleration is divided into two parts：relative dynamic state and steady state. The standard deviation, mean, energy and other characteristics are extracted from the parts. Then, these features which are selected from different walking speed are combined with the slope of the peak point line into a new feature set. Finally, the combination classifier is used to achieve better recognition accuracy. The experimental results show that the recognition of the walking in slow speed of the rate reaches 98.3% and walking in high speed reaches 97.6%. The method of this paper is better than the existing methods of personnel identification.

Key words: gait recognition, Hidden Markov, Model（HMM）, geographic division, acceleration sensor, connection slope of peak point, combined classifier

摘要： 为提高基于智能手机内置加速度传感器的人员识别率，提出了一种基于信息分割的组合分类器识别方法。根据人员步行加速度变化特点提出了基于HMM（隐马尔可夫模型）的划分方法，将人员步行加速度划分成相对动态与稳态两个部分，分别从两个区域提取标准差、均值、能量等特征；根据不同步行速率选择这些特征和峰值点连线斜率组合成新的特征集合；最后，采用组合分类器的方法获得了更加理想的识别精度。实验结果表明，在人员慢步行走的姿态下的识别率达到了98.3%，快速步行达到了97.6%。较现有人员识别方法有较大的提高。

关键词: 身份识别, 隐马尔可夫模型（HMM）, 区域划分, 加速度传感器, 峰值点连线斜率, 组合分类器