基于傅里叶描述子的手势轨迹模板匹配方法  

Gesture Trajectory Template Matching Based on Fourier Descriptors

作　　者：刘电霆[1] 梁桂宾 LIU Dianting;LIANG Guibin(College of Mechanical and Control Engineering,Guilin University of Technology,Guilin 541004,China)

机构地区：[1]桂林理工大学机械与控制工程学院,广西桂林541004

出　　处：《电子机械工程》2020年第1期1-5,共5页Electro-Mechanical Engineering

摘　　要：针对手机移动平台中人机交互的问题及手势识别中的时域问题,提出基于视觉的手势轨迹识别作为人与手机交互的方式。利用傅里叶描述子对手势轨迹进行特征描述,并结合置信度对动态手势轨迹进行分类。文中先对手势视频进行运动检测,利用基于颜色的粒子滤波器对手部进行跟踪以获取手势中心运动轨迹,对所得的运动轨迹进行闭合化处理,使得轨迹变成封闭曲线集,对封闭曲线集进行傅里叶变换,使得运动轨迹具有平移、缩放和旋转不变性,利用所得的傅里叶描述子与模板进行匹配,在一定范围内具有良好的稳定性。同时提出变形系数来描述采样所得运动轨迹的扭曲变形程度。实验结果表明该方法在变形系数为0.3以下时具有较高的准确性。In view of the man machine interaction in mobile phone platform and the time domain problem in gesture recognition,a visual based gesture trajectory recognition is proposed as a way of interaction between human and mobile phone.Fourier descriptors are used to characterize the gesture trajectory,and the dynamic gesture trajectory is classified combined with confidence coefficient.In this paper,the motion detection of ges ture video is firstly carried out,and the hand tracking is carried out by using the color based particle filter to obtain the trajectory of gesture center.The obtained trajectory is closed so that the trajectory becomes a closed curve set,and the closed curve set is Fourier transformed to make the trajectory have the invariance of transla tion,scaling and rotation.The obtained Fourier descriptors are matched with template and have good stability in a certain range.The deformation coefficient is proposed to describe the degree of distortion of the sampled trajectory.The experimental results show that the method has high accuracy when the deformation coefficient is less than 0.3.

关 键 词：手势轨迹识别 傅里叶描述子 变形系数 鲁棒性 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]