MANIFOLD WARP SEGMENTATION OF HUMAN ACTION

Abstract

Human action segmentation is important for human action analysis, which is a highly active research area. Most segmentation methods are based on clustering or numerical descriptors, which are only related to data, and consider no relationship between the data and physical characteristics of human actions. Physical characteristics of human motions are those that can be directly perceived by human beings, such as speed, acceleration, continuity, and so on, which are quite helpful in detecting human motion segment points. We propose a new physical-based descriptor of human action by curvature sequence warp space alignment (CSWSA) approach for sequence segmentation in this paper. Furthermore, time series-warp metric curvature segmentation method is constructed by the proposed descriptor and CSWSA. In our segmentation method, descriptor can express the changes of human actions, and CSWSA is an auxiliary method to give suggestions for segmentation. The experimental results show that our segmentation method is effective in both CMU human motion and video-based data sets.

Existing System

Gaussian mixture model (GMM), principle component analysis (PCA) and locality preserving projection (LPP).

Proposed System

In this paper, we assume that inconsistent frames are transition clips between different motions, and propose the Time Series-Warp Metric Curvature Segmentation (TS-WMCS) algorithm to realize human motion segmentation based on this assumption. TS-WMCS can adjust the segment points by analyzing temporal feature curves in low-dimensional space. The main contributions of this paper are as follows. 1) Describe motion continuity by local curvature like descriptors generated in angle space, and then utilize the descriptors to detect the transition points in motion sequences. 2) The local curve degree of the motion sequences described in angle space can reflect human kinetic features well. If human motion changes, the local curve degree of the motion sequences will change as well. 3) Propose a robust ID estimation algorithm based on a cosine metric (CM), and provide reliable IDs for human motion sequences. 4) Segment the motion sequence from multi perspectives by using temporal feature curves to describe the motion sequences in low-dimensional space, thereby improving the segmentation accuracy efficiently. 5) The TS-WMCS algorithm can guarantee the recall when dealing with simple motion sequences, and has a comprehensively better performance while segmenting complex motion sequences compared with other manifold-based segmentation algorithms.

Architecture

BLOCK DIAGRAM