Classification of Multi-class Trajectories Based on SVM
Libin Yang · Computer and Digital Engineering · 2012
Trajectory analysis is a significant way for the problem of anomaly detection in visual surveillance systems.This paper samples the trajectories and uses the sets of the coordinates as the feature vectors of the sample trajectories,then trains classifiers with SVM and realizes the classification of the multi-class trajectories.The experiment shows that our method can satisfy the requirement of the input data of the kernel functions in SVM,and can efficiently classifies the multi-class trajectories.