基于隐马尔可夫模型的3D手写识别方法

3D handwriting recognition method based on hidden Markov models

为实现便捷高效的人机交互,提高交互能力,在基于隐马尔可夫模型(HMM)的基础上,提出了一种新的3D手写识别方法。该方法使用带有三轴加速度传感器的手持设备去采集各种手写数据;使用插值及快速傅里叶变换(FFT)滤波等方法对采集的数据进行预处理;使用隐马尔可夫模型对每个手写动作进行模型训练;使用训练过的手写模型对采集的数据进行手写识别。数据测试结果表明,该方法在手持移动设备上数据分类的准确性可达到84.5%。

In order to achieve efficient and convenient human-computer interaction,improve interaction ability,this paper proposed a new 3D handwriting recognition method based on HMM.In this method,using of the tri-axis accelerometer mounted on a handheld device to collect various handwriting data;using of the interpolation and FFT filtering to preprocess the data;using of the hidden Markov models to train every handwriting movement;using of the trained HMM to carry out the handwriting recognition.Tested on the handwriting mobile device,the results show that the classification accuracy can reach 84.5%