基于Leap Motion和支持向量机的手势识别

为了使手势识别在更多的领域得到推广及应用,提出了基于Leap Motion体感设备实时跟踪技术获取手势三维空间坐标信息的方法,并从中分别提取角度信息和相对坐标信息,构建手势特征数据,建立手势识别模型.对特征数据进行归一化处理后,利用支持向量机(SVM)分类器进行训练、建模和分类,实现手势识别.实验结果表明:以角度数据和坐标数据作为手势特征的方法可行,平均识别率分别为96.6%和91.8%.通过对比可以得出:以角度数据作为特征值具有较高的准确性和鲁棒性,并避免了单纯依照一种特征值产生的局限性.

Hand gesture recognition based on Leap Motion and SVM

In order to make gesture recognition applied in more fields,an approach based on the Leap Motion somatosensory devices real-time tracking technology is put forward to obtain 3D space coordinate information of gesture,and extract angle information and relative coordinate information respectively,construct gesture feature data,establish gesture recognition model. After normalization of feature data,support vector machine(SVM) classifier is used for training,modeling and classification,and gesture recognition is realized. The experimental results show that the method which uses angle data and coordinate data as gesture feature is feasible,the average recognition rate is respectively 96.6% and 91.8%.It can be concluded that using angle data as feature value has higher accuracy and robustness,and it can avoid the limitation of a simple eigenvalue,it has strong convincing.