Section 1211 Ya thief Abstract

Section 1211 Ya thief Abstract
This report wood work addresses the technique of Video-based Human Motion Capture, gives a detail survey on the theories and techniques of Video-based Human Motion Capture and discusses the difficulties of motion capture and the possible solutions. This report also presents the work of tracking players from soccer match videos. Many tracking technique is used in this work, including image processing, mathematical morphology and Kalman filter to determine players' positions in the video sequence. A method is presented to implement wood work a quick segmentation and extraction of feature lines from a soccer video. The parameters of the white lines, which are around penalty area and goal area, are detected automatically. These parameters can be used to determine the camera parameters, reconstruct the soccer field, and compute the soccer players' real positions. In order to prepare the 'Technologically Advanced Olympiad' project, this report discusses a framework of motion analysis on weight lifting. Some tests plants are set up to try various motion capture techniques including a 3D stick model, camera calibration and qualitative analysis on human's silhouette.
Chapter 1 Introduction 1 1.1 video-based human motion capture technology and its applications a 1.2 Problem Description and general wood work tracking difficulty tracking framework 2 1.3 3 1.4 4 1.5 Introduction wood work future research trends of the previous system and the analysis made 4 1.5. 1 Pfinder system 5 1.5.2 twist motion and exponential maps based approach mistake! Bookmark not defined. 1.5.3 three-dimensional point matching method based on meta-sphere model error! Bookmark not defined. 1.5.4 deformation method based on block matching error! Bookmark not defined. Chapter 2 used motion tracking algorithm and key technical errors! Bookmark not defined. 2.1 motion tracking feature extraction and the underlying assumptions wrong! Bookmark wood work not defined. 2.2 based on motion prediction wood work mannequin and matching technical errors! Bookmark not defined. 2.3.1 mannequin mistake! Bookmark not defined. 2.3.2 Model-based matching technique mistake! Bookmark not defined. 2.3 Least Squares Error! Bookmark not defined. 2.4 Kalman filtering technique mistake! Bookmark not defined. About 2.5 EM method mistake! Bookmark not defined. Chapter 3 video game football player tracking error! Undefined Bookmarks. 3.1 division stands, the players and the white line error! Bookmark not defined. 3.1.1 extract bleachers and the players mistake! Bookmark not defined. wood work 3.1.2 white line dividing wrong! Bookmark not defined. 3.2 feature extraction line parameter error stadium! Bookmark not defined. Principle 3.2.1 Hough transform mistake! Bookmark not defined. 3.2.2 extract linear parameter error! Bookmark not defined. 3.3 player tracking error! Bookmark not defined. 3.4 The results, problems and improvement plans mistake! wood work Bookmark not defined. Preparations Chapter 4, "Technology Olympics" wood work of the project and future research in the wrong direction! Bookmark not defined. 4.1 Establishment and preliminary results of the experimental environment mistake! Bookmark not defined. 4.2 Olympic motion capture technology projects to achieve the vision mistake! Bookmark not defined. 4.3 Future work planning and research directions wrong! Bookmark not defined. 4.3.1 back to the main work so wrong! Bookmark not defined. 4.3.2 Future Work Plan mistake! Bookmark not defined. Acknowledgements References 7 8 Chapter 1 Introduction video of human motion tracking, automatic identification and interpretation of human behavior has always had many important applications, such as automated surveillance wood work systems, virtual wood work reality, human-machine interface, sports and Analysis of sports medicine and so on. Wherein the components wood work of the human body and the tracking of the whole system is the most important part, identify and analyze motion shall be based on it. After 20 years of development, the current video-based tracking technology combines image processing, theoretical computer vision, computer graphics, artificial intelligence, and kinesiology and other subjects, became a popular multi-disciplinary fields. This chapter first reviews the classification based on video tracking technology and its application in different areas of the background, and then summarize wood work some of the more mature wood work the current framework, while difficulties and current wood work research trends in this area of research analysis. Finally, the more successful tracking system, as well as some of its domestic and analysis of these systems. 1.1 video-based human motion capture technology and its application to human motion capture means "process, in some resolution, capturing large-scale human movement" [Gavrila1997], referred to here is the large-scale movement of fingers upper torso limb exclude expressions and gestures and other small-scale wood work action. The market is relatively mature human motion capture system based on electric machines (Electromechanical), electromagnetic (Electromagnetic) and special optical mark (Retro- reflective Marker) and other types. Magnetic or optical tag is attached on the human body, their three-dimensional motion trajectories are used to describe the target, these systems are automated, but the equipment is very bulky, and expensive. With the popularity of low-cost digital cameras and high-performance home PC's, video-based human motion capture is becoming wood work a hot topic.
Video-based human motion capture many potential applications, including automated surveillance, human-computer interaction interface, human motion analysis, animation, film and television special effects production, content-based video encoding wood work and so on. In the application of automatic monitoring, the system for single or multiple target tracking, and analyzing its behavior according to the motion characteristics wood work of the target. This system wood work can be applied to monitor the parking lot, supermarkets, ATMs and other places. In this behind wood work the football game player tracking can also be attributed to this area. Tracking technology can also be used as a computer to interact in a way, through the human head and limb to get two-dimensional tracking of each part of the body or change in position three-dimensional space, and accordingly recognize the body's movement behavior, then drive or control other applications. Here, the tracking wood work technology can be used sign language recognition, game interface, virtual reality, and remote control, remote meetings. For example: wood work In the MIT Media Lab STIVE systems, human tracking technology combined with virtual reality is used to help cancer patients. Patients by tai chi to control virtual reality to overcome cancer of white blood cells, increase the confidence of patients fight cancer [Becker1996]. Human motion analysis is mainly used in the medical and sports fields. In medicine, we can use motion analysis of the patient gait analysis. In sports, you can analyze the movement of athletes to help them improve wood work motor skills.
1.2 Description of the problem and the general framework for the calculation according to tracking Marr's theory of vision, vision is an information processing wood work system can be divided into computational theory, and algorithms that implement the three levels of hardware. In answer to the purpose of calculating the theoretical level (input and output) and the strategy of the visual system. The task is to study how the visual system to build relationships between the input and output constraints, three-dimensional information of the object (shape, position and attitude) How to recover from a two-dimensional gray-scale images. On the indicator with the algorithm level answered how to represent input and output information, wood work how to implement the function corresponding wood work to the theoretical calculations of the algorithm, and how to convert from one representation to another representation [TRANSACTIONS 2000]. According to Marr's theory can be seen as a search for correspondence between the video and the process of human motion tracking system based on the video of the human body. Enter one or more of the perspective of the image sequence, the output is related wood work to human movement motion parameters. Depending on the input tracking system, the output is different, so their representation and achieve different. Depending on the assumptions and methods used on the human body model and the low-level features can be used for this tracking problem have different classification. Such as model-based and not based on the model, a multi-camera and single-camera and so on. Marr visual system is divided into three stages from bottom to top, the vision he described as a gradual process characterized by a low-level abstraction for the advanced features of the process [TRANSACTIONS 2000]. However, in human tracking this particular issue, there are a lot of prior knowledge can be used, including human skeleton model and human movement model. So most of the system uses a model based on human movement tracking. By constraining constraints and motion model skeleton model to guide higher body matches the description of the process from the ground motion parameters characteristic to.
Literature [Gavrila1997] The human motion tracking process is divided into the following four phases: initialization (initialization), wood work tracking (tracking), pose estimation (pose estimation), recognition (recognition). 1. Initialization process includes camera calibration, obtaining environmental characteristics, background model, the initial parameter mannequin wood work (bone length) and the model initial attitude, generally handmade participation. 2. In the tracking phase, the use of image segmentation to obtain low-level features, characteristics and conduct correspondence between frames. 3. In the pose estimation wood work stage, low-level features are matched to the human body model, whereby the attitude of people in the current frame. 4. In the recognition phase, wood work the system by analyzing the result, the behavior of the state of human motion. This process wood work is not a simple wood work relationship between the four serial, intermediate results of each stage can be fed back to the previous stage, in front of the guidance process. This article will focus on the first twenty-three research stage, for the identification process does not make too much discussion. 1.3 Difficulties in tracking problem in computer vision even after decades of research has made great progress, but because people lack the human visual mechanism particularly in relation to the human aspects of intelligent understanding of the mechanism, computer vision and human vision there is considerable gap. The general said was machine vision only works under controlled conditions. The problem is a lack of knowledge about the computer environment, and other aspects of the human body. If you are not in the controlled conditions, the computer does not even exist not determine whether someone from a single photograph. So to avoid this classic problem in computer vision problems in tracking method is to fully utilize the knowledge about the human body, and with appropriate restrictions wood work assumptions, integrated multi-tracking method. Specific to human tracking this issue, deal with the difficulties involved are the following: 1. Under the assumed conditions are not narrated, automatic separation of human and background. This is actually a division problem, the problem is still not a good solution. wood work 2. Deal with occlusion occlusion between each arm and body. Many current systems are assumed to be tracked target is not blocked when occlusion wood work occurs, often because of the loss of target tracking and failure. The current solution comprising: when the occurrence of occlusion, calculated using data before and after the frame is blocked by the target position. Using multi-camera, a video camera when the camera wood work fails with other data to track. 3. If the limbs and body textures same clothes, due to lack of contrast, when the arm and body intersect, information can not be isolated from the arm. 4. How to achieve without human intervention automatic initialization and automatically recover from errors. Prerequisite many motion capture system that runs in the first frame of the video manually initialize wood work mannequins and posture, but this method wood work requires wood work not only people's work, but also because of the manual initialization information will gradually weaken over time, it is impossible to avoid the accumulation wood work errors and other disturbances. 5. The cumulative impact of handling errors, and illumination changes and other environmental factors, to achieve tracking movement (as opposed to a few seconds of motion) long. Because of the above difficulties, according to the reading of literature, currently does not have any vision of a computer-based system capable of tracking