Dissertation > Excellent graduate degree dissertation topics show

Study and Implementation of Human Action Recognition Based on Semi-Connected HMM

Author: FuXiaoYing
Tutor: XuDe
School: Beijing Jiaotong University
Course: Computer Science and Technology
Keywords: Moving image processing Recognition of human action Human motion characteristics HMM BME-SCHMM
CLC: TP391.41
Type: Master's thesis
Year: 2009
Downloads: 308
Quote: 2
Read: Download Dissertation

Abstract


Human behavior recognition is an important branch of the field of artificial intelligence , widespread concern in recent years by researchers in the field of machine vision , and related scientific research is bringing a new interactive way to people's lives . Human behavior recognition basic steps are: 1. Motion detection and tracking purpose is related to the motion feature extracted from the video sequence ; human posture estimation and identification, the purpose is to create a motion model based on the motion characteristics , thereby performing identification of human behavior . In view of the basis of previous studies , the paper mainly studies the second step . Far, the hidden Markov model (HMM) used in human behavior recognition mostly used is fully connected model HMM (FCHMM) , to obtain a more satisfactory recognition result . However, these methods only subjectively selected number of states and mechanically using the structure of a full connection , without considering the relationship between the human behavior and HMM layout . To address this issue , the use of a new semijoin before the tri-state the HMM (BME-SCHMM) of human behavior recognition , the combination of human behavioral characteristics and the design of the HMM parameters , by limiting the number and status of the transfer conditions to improve the accuracy and reduce the computational complexity of human behavior recognition . Specific algorithm implementation process , the output probability for each state , the introduction of the concept of weight . Israeli the Weizmann human behavior database and the Swedish KTH human behavior database test , with the traditional fully connected HMM human behavior recognition model comparison , concluded : this method can get similar recognition rate can be reduce the number of states and limits the state transition conditions , and thereby improves the computing performance . Low complexity of the algorithm , fast and effective , with good versatility , has a better performance than the existing human recognition algorithm .

Related Dissertations

  1. Multiple ANN/HMM Hybrid Used in Speech Recognition,TN912.34
  2. Measurement and Reconstruction of Ion Channel Current for Cell Membrane in Rat’s Cerebral Cortex,Q42
  3. HMM-based Connection between the Social Network Analysis,F49
  4. Complex environment speech enhancement method,TN912.35
  5. Research on Human Abnormal Behavior Detection and Recognition in Intelligent Video Surveillance,TP391.41
  6. Research on Hmm-based Speech Recognition System of the Robot,TN912.34
  7. The Study of Cough Signal Recognition Based on HMM-ANN Hybrid Model,TN912.34
  8. Hidden Group Detection Based on Evolution of Hidden Markov Model,O157.5
  9. State and Parameter Estimation of HMMs with Transition Density Function,O211.62
  10. Research of Multi-Sensory Myoelectric Prosthetic Hand with Hardness and Thermal Conductivity,TP242
  11. Research and Application of Three-Dimensional of Human-Computer Interaction Technology Based on Computer Vision,TP391.41
  12. Automatic Chord Recognition Based on Hidden Markov Models,TP391.3
  13. Study on Hybrid Model of Speech Recognition Based on HMM and PNN,TN912.34
  14. Research on the Key Technologies fo Speech Recognition for Robot Communication,TN912.34
  15. Analysis of the simulation of the risk of credit card fraud detection system based on HMM,F832.2
  16. The Research of Keystroke Dynamics Based on Statistical Method,TP393.08
  17. Analyzing and Predicting the Run-time Behavior of Software in Electronic Trading,TP311.52
  18. Study of Recognitizing Printed Chinese Characters on Packaging Boxes,TP391.41
  19. The Research of Sign Language Recognition Technology Based on Complex Background,TP391.41
  20. Research of Expression Recognition Based on EHMM,TP391.41
  21. Speech Emotion Recogonition Research,TN912.34

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net  Mobile