Recognizing human actions from video sequences using invariant shape

In this paper, recognizing human actions has been investigated from video sequences. With morphological gradient and pooling operation, invariant shape of each human body in an action sequence is obtained. Edge feature of invariant shape is extracted to represent human actions. Pyramid Histograms of Orientation Gradients (PHOG) of all invariant shapes in video are averaged to form a feature vector that captures the characteristic of human actions in this video sequence. Using Support Vector Machine (SVM), the method is tested on the KTH action dataset. The obtained impressive results show that invariant shape is more effective than original video in human action recognition.