Human Action Recognition in Video Forms

From various scattered similar the action is recognized on the basis of space time and volume in video forms, which can be fully characterized by a linear rank decomposition. Recurrence plot theory is applied, and introducing the concept of Joint Self-Similarity Volume (Joint-SSV) to model this scattered action in various forms, and hence rank-1tensor approximation of the Joint-SSV is applied to obtain very low-dimensional descriptors that very correctly characterize an action in a video sequence id used [5].The descriptor vectors make it more possible to recognize actions without explicitly aligning the videos in time in order to compensate for speed of execution or differences in video frame rates. The method is generic, in the sense that it can be applied using different level features, such as tracked points, 2D block diagrams, histogram of oriented gradients. Therefore, the method does not necessarily require explicit tracking of features in the space-time volume.