A KPCA spatio-temporal differential geometric trajectory cloud classifier for recognizing human actions in a CBVR system
作者：
Highlights：
• A CBVR based upon spatio-temporal trajectories of human motion.
• Trajectories encode two motion scales: large (hyperplanes); small (local differential geometry).
• New fuzzy KNN classifies trajectories with proximity, orientation and approximate membership.
• Validation of video retrieval system from standard datasets and feature length films.
摘要
•A CBVR based upon spatio-temporal trajectories of human motion.•Trajectories encode two motion scales: large (hyperplanes); small (local differential geometry).•New fuzzy KNN classifies trajectories with proximity, orientation and approximate membership.•Validation of video retrieval system from standard datasets and feature length films.
论文关键词：Human motion recognition,Content based video retrieval,Spatio-temporal templates,Kernel-PCA,Frenet–Serret formulas,Differential curvature,Fuzzy-KNN
论文评审过程：Available online 28 March 2015.
论文官网地址：https://doi.org/10.1016/j.eswa.2015.03.010
