人机交互动态手势轮廓提取仿真研究
|
摘要
为提高动态手势识别的精确性,在手势特征提取、动态手势识别等方面进行实验研究。采用基于肤色模型的分割方法对手势图像进行处理,分别对手势轮廓特征和手势运动特征进行提取,提出基于HMM-NBC模型的手势识别算法,定义10种手势并建立动态手势样本库,进行手势识别研究,并与支持向量机手势识别算法相比较,研究表明:HMM-NBC模型算法的手势识别速度明显高于支持向量机算法,具有较高的识别率,平均手势识别率为88.8%。
In order to improve the accuracy of dynamic gesture recognition,experimental research was conducted on gesture feature extraction and dynamic gesture recognition. The gesture image is processed by the skin color-based segmentation method. Gesture outline features and gesture movement features are extracted respectively. A gesture recognition algorithm based on HMM-NBC model is proposed,10 gestures are defined and a dynamic gesture sample database is established for gesture recognition. Compared with the support vector machine gesture recognition algorithm,the research shows that:HMMNBC model algorithm gesture recognition speed is significantly higher than the support vector machine algorithm,with a high recognition rate,the average gesture recognition rate of 88.8%.
In order to improve the accuracy of dynamic gesture recognition,experimental research was conducted on gesture feature extraction and dynamic gesture recognition. The gesture image is processed by the skin color-based segmentation method. Gesture outline features and gesture movement features are extracted respectively. A gesture recognition algorithm based on HMM-NBC model is proposed,10 gestures are defined and a dynamic gesture sample database is established for gesture recognition. Compared with the support vector machine gesture recognition algorithm,the research shows that:HMMNBC model algorithm gesture recognition speed is significantly higher than the support vector machine algorithm,with a high recognition rate,the average gesture recognition rate of 88.8%.
引文
[1]Zhang Qiu-Yu,Lu Jun-chi,Wei Hui-yi.Dynamic hand gesture trajectory recognition based on block feature and skin-color clustering[J].International Journal of Multimedia&Ubiquitous Engineering,2016:21-34.
[2]郭晓利,杨婷婷,张雅超.基于Kinect深度信息的动态手势识别[J].东北电力大学学报,2016,36(2):90-94.(Guo Xiap-li,Yang Ting-ting,Zhang Ya Chao.Gesture recognition based on kinect depth information[J].Journal of Northeast Dianli University,2016,36(2):90-94.)
[3]庞海波,李占波,丁友东.基于时间序列手势轮廓模型的动态手势识别[J].华南理工大学学报:自然科学版,2015,43(1):140-146.(Pang Hai-Bo,Li Zhan-Bo,Ding You-Dong.Dynamic gesture recognition based on gesture contour model of time series[J].Journal of South China University of Technology,2015,43(1):140-146.)
[4]王真水,李琳,刘晓平.支持自定义的多粒度动态手势识别方法研究[J].电子测量与仪器学报,2014,28(4):416-423.(Wang Zhen-shui,Li Lin,Liu Xiao-ping.Research on multi-granularity dynamic gesture recognition method of supporting self-definition[J].Journal of Electronic Measurement&Instrumentation,2014,28(4):416-423.)
[5]何超,胡章芳,王艳.一种基于改进DTW算法的动态手势识别方法[J].数字通信,2013,40(3):21-25.(He Chao,Hu Zhang-fang,Wang Yan.Novel dynamic gesture recognition method based on improved DTW algorithm[J].Digital Communication,2013,40(3):21-25.)
[6]李文生,解梅,邓春健.基于Hermite神经网络的动态手势学习和识别[J].计算机工程与科学,2012,34(2):116-122.(Li Wen-sheng,Xie Mei,Deng Chun-Jian.Dynamic gesture learning and recognition based on the hermite neural network[J].Computer Engineering&Science,2012,34(2):116-122.)
[7]Karwowski J,Okulewicz M,Legierski J.Application of particle swarm optimization algorithm to neural network training process in the localization of the mobile terminal[J].Communications in Computer&Information Science,2013(383):122-131.
[8]宋红,余隋怀,陈登凯.基于深度图像技术的可供性物体获取方法[J].机械设计与制造,2015(3):28-31.(Song Hong,Yu Sun-huai,Chen Deng-kai.An affordance object acquiring method based on depth image technology[J].Machinery Design&Manufacture,2015(3):28-31.)
[2]郭晓利,杨婷婷,张雅超.基于Kinect深度信息的动态手势识别[J].东北电力大学学报,2016,36(2):90-94.(Guo Xiap-li,Yang Ting-ting,Zhang Ya Chao.Gesture recognition based on kinect depth information[J].Journal of Northeast Dianli University,2016,36(2):90-94.)
[3]庞海波,李占波,丁友东.基于时间序列手势轮廓模型的动态手势识别[J].华南理工大学学报:自然科学版,2015,43(1):140-146.(Pang Hai-Bo,Li Zhan-Bo,Ding You-Dong.Dynamic gesture recognition based on gesture contour model of time series[J].Journal of South China University of Technology,2015,43(1):140-146.)
[4]王真水,李琳,刘晓平.支持自定义的多粒度动态手势识别方法研究[J].电子测量与仪器学报,2014,28(4):416-423.(Wang Zhen-shui,Li Lin,Liu Xiao-ping.Research on multi-granularity dynamic gesture recognition method of supporting self-definition[J].Journal of Electronic Measurement&Instrumentation,2014,28(4):416-423.)
[5]何超,胡章芳,王艳.一种基于改进DTW算法的动态手势识别方法[J].数字通信,2013,40(3):21-25.(He Chao,Hu Zhang-fang,Wang Yan.Novel dynamic gesture recognition method based on improved DTW algorithm[J].Digital Communication,2013,40(3):21-25.)
[6]李文生,解梅,邓春健.基于Hermite神经网络的动态手势学习和识别[J].计算机工程与科学,2012,34(2):116-122.(Li Wen-sheng,Xie Mei,Deng Chun-Jian.Dynamic gesture learning and recognition based on the hermite neural network[J].Computer Engineering&Science,2012,34(2):116-122.)
[7]Karwowski J,Okulewicz M,Legierski J.Application of particle swarm optimization algorithm to neural network training process in the localization of the mobile terminal[J].Communications in Computer&Information Science,2013(383):122-131.
[8]宋红,余隋怀,陈登凯.基于深度图像技术的可供性物体获取方法[J].机械设计与制造,2015(3):28-31.(Song Hong,Yu Sun-huai,Chen Deng-kai.An affordance object acquiring method based on depth image technology[J].Machinery Design&Manufacture,2015(3):28-31.)