由于基于图像处理的手势识别方法对环境背景要求较高且存在不稳定性问题,文章使用三维加速度传感器的连续数据进行手势识别.三维加速度传感器内置于大部分智能手机中,具有应用方便的特点.实验通过传感器获取加速度信号,经过低通滤波、去重力和特征提取的信号预处理过程后,结合隐马尔可夫模型和混合高斯模型的理论方法,实现手机手势的连续识别,并驱动应用层预先定义的交互命令.%Gesture recognition based on image processing method depends highly on the environment background,and produces inconsistent results. A new gesture recognition method is designed based on 3D accelerometers installed on smart phones, supporting gesture based interaction for mobile environments. The acceleration signal is captured, and filter off the noises by means of low pass filtering,gravity cut, and feature extraction. Combined with the Hidden Markov Model and Gaussian Mixture Model, continuous gestures recognition of smart phones is implemented to drive the corresponding functions predefined on applications level. The Experiment results demonstrate that the gestures recognition has higher rate of accuracy.
展开▼
