作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2018, Vol. 44 ›› Issue (12): 271-275,287. doi: 10.19678/j.issn.1000-3428.0049178

• 图形图像处理 • 上一篇    下一篇

有效视频帧时间序池化的人体行为识别算法

鹿天然,于凤芹,陈莹   

  1. 江南大学 物联网工程学院,江苏 无锡 214122
  • 收稿日期:2017-11-03 出版日期:2018-12-15 发布日期:2018-12-15
  • 作者简介:鹿天然(1993—),女,硕士研究生,主研方向为模式识别、智能系统;于凤芹、陈莹,教授、博士。
  • 基金资助:

    国家自然科学基金(61573168);中央高校基本科研业务费专项资金(JUSRP51733B)。

Human Action Recognition Algorithm with Temporal Rank Pooling of Valid Video Frames

LU Tianran,YU Fengqin,CHEN Ying   

  1. School of Internet of Things Engineering,Jiangnan University,Wuxi,Jiangsu 214122,China
  • Received:2017-11-03 Online:2018-12-15 Published:2018-12-15

摘要:

为利用人体行为的时域信息并减少帧间冗余及特征维数,提出一种提取有效视频帧并对其时间序池化的人体行为识别算法。通过对视频帧的稠密轨迹特征进行局部累计描述向量编码,获取视频帧特征表示,对每帧的特征编码进行余弦相似度分析,剔除冗余特征帧得到有效视频帧特征序列。采用时间序池化对有效视频帧特征序列进行排序,得到可表示视频时序动态变化的特征向量,然后训练支持向量机实现人体行为识别。在HMDB51和UCF101数据集上的实验结果表明,与稠密轨迹行为识别算法相比,该算法可有效提高识别准确率。

关键词: 行为识别, 稠密轨迹, 局部累计描述向量, 余弦相似度分析, 时间序池化

Abstract:

In order to make full use of the video-wide temporal information and reduce the redundant frames and dimensions of features,a method of extracting valid video frames and performing temporal rank pooling for human action recognition is proposed.Vector of Locally Aggregated Descriptors(VLAD) is used to encode dense trajectory features of every frame of video to get feature representations.The cosine similarity analysis of frame features is employed to remove the redundant features and extract feature sequence of valid video frames.Temporal rank pooling is performed to order feature sequence of valid frames temporally and get the feature vectors capturing the evolution of video-wide temporal information.Support Vector Machine(SVM) is learned to get the results of human action recognition.Experimental results conducted on HMDB51 and UCF101 datasets show that compared with the dense trajectory recognition algorithm,the proposed agorithm has improved the recognition accuracy.

Key words: action recognition, dense trajectory, Vector of Locally Aggregated Descriptors(VLAD), cosine similarity analysis, temporal rank pooling

中图分类号: