Human Action Recognition Using Depthmap Image Sequences for Abnormal Event Detection<br/>

Please enable javascript in your browser.

Human Action Recognition Using Depthmap Image Sequences for Abnormal Event Detection

1406 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 49038 (05)
University: Sharif University of Technology
Department: Electrical Engineering
Advisor(s): Mohammadzadeh, Hoda
Abstract:
The human action recognition is one of the most important concepts of computer vision in recent decades. Most of the two dimensional methods in this field are facing serious challenges such as occlusion and missing the third dimension of data. Development of depth sensors has made easy access to tracking people and 3D positions of human body joints. This Thesis proposes a new method of action recognition that utilizes the position of joints obtained by Kinect sensor. The learning stage uses Fisher Linear Discriminant Analysis (LDA) to construct discriminant feature space. Two types of distances, i.e., Euclidean and Mahalanobis, are used for recognizing the states. Also, Hidden Markov Model (HMM), has been used to recognize the action related to each sequence of states and therefore the actions are classified. According to results, this method significantly outperforms other popular methods, with recognition rate of 88.26% for eight different actions and up to 95.42% for classifying fall actions
Keywords:
Human Action Recognition ; Falling ; Hidden Markov Model ; Kinect Sensor ; Linear Discriminant Analysis ; Abnormal Activity ; Three Dimentional Joint Position