[1]宋相法,姚旭.基于多特征的深度图像序列人体行为识别[J].计算机技术与发展,2018,28(06):30-34.[doi:10.3969/ j. issn.1673-629X.2018.06.007]
　SONG Xiang-fa,YAO Xu.Human Activity Recognition from Depth Image Sequences Based on Multiple Features[J].,2018,28(06):30-34.[doi:10.3969/ j. issn.1673-629X.2018.06.007]
点击复制

基于多特征的深度图像序列人体行为识别()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
28
期数:
2018年06期
页码:
30-34
栏目:
智能、算法、系统工程
出版日期:
2018-06-10

文章信息/Info

Title:
Human Activity Recognition from Depth Image Sequences Based on Multiple Features
文章编号:
1673-629X(2018)06-0030-05
作者:
宋相法; 姚旭
河南大学 计算机与信息工程学院,河南 开封 475004
Author(s):
SONG Xiang-fa; YAO Xu
School of Computer and Information Engineering,Henan University,Kaifeng 475004,China
关键词:
人体行为识别; 深度图像序列; 多特征; 核极限学习机
Keywords:
human activity recognition; depth image sequences; multiple featured; kernel extreme learning machine
分类号:
TP301.6
DOI:
10.3969/ j. issn.1673-629X.2018.06.007
文献标志码:
A
摘要:
由于现有方法多采用单一特征研究深度图像序列人体行为识别,其识别性能较低。 针对上述问题,提出了一种基于超法向量特征与深度运动图的梯度方向直方图特征的多特征行为识别方法。该方法首先从深度图像序列中提取两种特征:超法向量特征和深度运动图的梯度方向直方图特征,以增强特征互补性;然后利用核极限学习机分别获得两种特征的识别结果;最后对识别结果利用对数意见汇集规则进行融合得到最终识别结果。 在 MSR Action3D 数据集上进行了测试,得到了 96.3%的识别率,不但超过了基于超法向量特征方法的识别率和基于深度运动图的梯度方向直方图特征方法的识别率,而且也超过了其他方法的识别率。
Abstract:
he current methods cannot yield satisfactory performance since they just employ single feature for research of human activity recognition from depth image sequences. For this,we propose a novel method based on multiple features including the super normal vector features and the histograms of oriented gradients features from the depth motion maps. Firstly,in order to increase the complementarities of features,we extract two types of features from depth image sequences including the super normal vector features and the histograms of oriented gradients features from the depth motion maps. Then we use the kernel extreme learning machine to obtain the recognition result of two features. Finally,the rule of logarithmic opinion pool is used to combine the classification outcomes. The tests on MSR Action3D dataset show that it can achieve recognition ratio of 96.3% which is higher not only than that of the methods based on the super normal vector features and based on the histograms of oriented gradients features from the depth motion maps,but also than that of other methods.

参考文献/References:

 

相似文献/References:

[1]宋相法,姚旭.基于多描述子特征编码的人体行为识别[J].计算机技术与发展,2018,28(08):17.[doi:10.3969/ j. issn.1673-629X.2018.08.004]
　SONG Xiang-fa,YAO Xu.Human Activity Recognition Based on Multi-descriptor Feature Coding[J].,2018,28(06):17.[doi:10.3969/ j. issn.1673-629X.2018.08.004]
[2]朱连章,陈殿明,郭加树,等.基于协同LSTM神经网络的人体行为识别研究[J].计算机技术与发展,2018,28(12):79.[doi:10．3969/j． issn．1673－629X．2018．12．017]
　ZHU Lianzhang,CHEN Dianming,GUO Jiashu,et al.Research on Human Action Recognition Based on Synergistic LSTM Neural Network[J].,2018,28(06):79.[doi:10．3969/j． issn．1673－629X．2018．12．017]
[3]宋相法,吕 明.融合三维骨架和深度图像特征的人体行为识别[J].计算机技术与发展,2019,29(07):55.[doi:10. 3969 / j. issn. 1673-629X. 2019. 07. 011]
　SONG Xiang-fa,LYU Ming.Human Activity Recognition Based on Fusing 3D Skeleton and Depth Image Feature[J].,2019,29(06):55.[doi:10. 3969 / j. issn. 1673-629X. 2019. 07. 011]

更新日期/Last Update: 2018-08-16