Dissertation > Excellent graduate degree dissertation topics show
Study on Model-based Human Motion Tracking and Gesture Analysis Technology
Author: LiChangYan
Tutor: GuoLiJun
School: Ningbo University
Course: Applied Computer Technology
Keywords: HOG feature Kalman filter tracking Graph Model Human pose estimation
CLC: TP391.41
Type: Master's thesis
Year: 2011
Downloads: 80
Quote: 0
Read: Download Dissertation
Abstract
|
The video images of the human body tracking and pose estimation is a hot field of computer vision problems , and has broad application prospects . Currently existing methods have certain shortcomings, such as : attitude estimation inaccurate and slow, so studying a good tracking and pose estimation approach is very necessary . Variety of human characteristics , human motion analysis in feature selection and extraction is the key issue . Through correlation analysis , the paper used in the human body detection phase is characterized by the gradient orientation histogram ; Because the human body tracking and detection is the prerequisite for pose estimation , test results have a direct impact tracking and pose estimation results , the original method is difficult to achieve the effect of time , this be improved in the calculation of the introduction of integral HOG feature vector diagram HOG feature can effectively reduce the computation time . For human motion tracking methods , depending on the required application and target tracking features to choose different methods . This requires real-time performance and prospects for more accurate observation , so easy to implement using the Kalman filter -based tracking method . This innovation will HOG and Kalman together, not only can accurately track the human target location , but also can determine the size of the human body , in order to lay a good foundation for pose estimation . For human pose estimation , we use a method based on graph structure model , and its improvements. In order to reduce the search space , before making pose estimation , carried out some prelude treatment: first conducted a human detection , detection through the body to determine the general location and scale of the human body , the detection window for pose estimation of the input , can greatly reduce the search space and improve pose estimation speed ; detection window while adding some restrictions based on a priori knowledge , such as: human head and torso are generally located in the middle of the detection window , head in the middle of the upper torso , trunk, located just below the head , so that further reducing the search space and improve the speed. The body of the video sequence for attitude estimation, this not only utilizes the apparent continuity between video frames , but also use the geometric continuity , so that not only can improve the body pose estimation can improve the estimation accuracy of speed.
|
Related Dissertations
- Hysteresis -based optimization of vehicle routing problem,O224
- Overall integration of vehicles and components feature detection method,TP391.41
- Graph model based on statistical and machine learning algorithm and its application of certain,TP181
- Research on Extended Role Graph Model Based on Hybrid Hierarchy,TP393.08
- Text Categorization of Chinee Small Sample Based on Graph Model,TP391.1
- Query Conceptual Graph Guided Web Search Result Analysis,TP391.3
- Research on Face Recognition System in Video,TP391.41
- Deadlock detection based on advanced production scheduling of manufacturing systems,TH165
- Image Segmentation Based on Graph Models,TP391.41
- Generated based on random graph model of emotion,TP391.41
- Basic Research and Development for Municipal Drainage Design Software Based on Graph Theory,TP311.52
- Research of Product Opinion Mining Algorithm for Web Texts,TP391.1
- Study on Clustering and Classification with Applications,TP18
- KL distance - based Bayesian network structure learning algorithms,TP183
- Study on Network-Constrained Detection Algorithm Based on Graphical Model,TN929.5
- Hierarchical Topic Model Based on Document Frequency,TP391.1
- Routing Design in Cognitive Wireless Mesh Network Using Layered Graph and Ant Colony Algorithm,TN929.5
- Research of Image Analysis Based on Graph Model,TP391.41
- The Research on Prediction Model and Prediction Method of Short-term Traffic Flow,TP399-C6
- Sensing object segmentation method,TP391.41
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
|