High Impact Factor : 4.396 icon | Submit Manuscript Online icon |

Modeling and Detecting Human Action using Animated Pose Templates

Author(s):

Keerthika S , nsn college of engineering and technology; Kirubakaran.B, nsn college of engineering and technology

Keywords:

HOG-Histrogram of gradiant, HOF-Histrogram of optical flow, HMM-Hidden markov Model

Abstract

This research paper pose templates are used to identify short term, long term, and contextual action from a video scene. Each pose template consists of two components: 1) shape template represented by Histogram of Oriental Gradient (HoG). 2) The motion template represented by Histogram of Optical-Flows (HoF). Totally five videos are taken in training phase. The images are converted into frames. Neural Network is used to identify these actions from the videos. The images are divided into back view and front view. The back view image is subtracted from the original image. Then from the front view image we extract the actions of the person. The HoG is based on identifying the edges of the image. The HoF is based on threshold value. The threshold value is the grade value. A shape template may have more than one motion template represented by or -node. Therefore, each action is defined as a mixture (Or-node) of pose templates in an and-Or tree structure. While this pose template is apposite for detecting short-term action snippets in two to five frames, we spread it in two ways: 1) For long-term actions, we activate the pose templates by adding temporal constraints in a Hidden Markov Model (HMM), and 2) for contextual actions which are detected by the complete set by using SIMULATION and area is obtained by using the MATLAB 2013 software.

Other Details

Paper ID: IJSRDV4I90291
Published in: Volume : 4, Issue : 9
Publication Date: 01/12/2016
Page(s): 945-949

Article Preview

Download Article