Using Dynamic Time Warping Algorithm Optimization For Fast Human Action Recognition
Journal Title: Acta Technica Napocensis- Electronica-Telecomunicatii (Electronics and Telecommunications) - Year 2010, Vol 51, Issue 2
Abstract
In this paper, we present an approach based on dynamic programming and neural network for recognition and matching human action. Each human action is represented by the angular motion of the body parts. Each body part angular motion is represented by an one-dimensional time series. These time series are then compared separately for every body part with templates using dynamic programming (DTW). The results of the comparisons are used as input for a neural network that classifies the human action.
Authors and Affiliations
Tamás VAJDA
Keywords
Related Articles
- EP ID EP113313
- DOI -
- Views 117
- Downloads 0
FPGA Implementation of LMS and NLMS adaptive filters for acoustic echo cancellation
The paper proposes a register transfer level (RTL) description of two well-known adaptive algorithms used in acoustic echo cancellation: the Least Mean Square (LMS) and Normalized Least Mean Square (NLMS). The RTL descri...
Evaluating the Performances of the CastGate Tunnel Server over TCP and UDP Links in Multi-Client Configuration
The CastGate architecture defines a transitory tunneling solution, offering access to native multicast. This paper describes the implementation of a testing application (CastGate Tester - CT) for the CastGate Tunnel Ser...
Predicting Churn In Mobile Telecommunications Industry
In today’s competitive market, the customer retention represents one of the focal concern for telecommunications companies. The telecommunications sector can be considered on the top of the list with an annual churn rat...
Camera Phone Based Barcode Decoding System
We propose a 1D barcode acquisition and decoding system based on the use of J2ME enabled mobile phones. The approach relies on image processing techniques to correct the distortions introduces by the acquisition device,...
Fiber Segmentation in Composite Materials Using Marked Point Processes
This paper presents a new method dedicated to unsupervised 2D segmentation of fibers in a section of composite carbon-fiber materials. The framework relies on a marked point process algorithm. We shall create random elli...