Human tracking for computer virtual reality applications

• 主要作者: Teo, Han Seng.
• 其他作者: Shen Zhiqi
• 格式: Final Year Project
• 語言: English
• 出版: 2010
• 主題: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
• 在線閱讀: http://hdl.handle.net/10356/35240
• 機構: Nanyang Technological University

Computer vision is an interdisciplinary field related to many fields involving AI (Artificial Intelligence), robotics, geometry, computer programming, machine learning, signal processing etc. It is aggregations of ideologies from various fields with the purpose of creating a computer program or function to understand features in an image. In this Final Year Project, many techniques were explored for the purpose of developing the programming functions in a computer programming language Visual C++ to satisfy the milestones which were specified in “introduction” section. The report presents a unique process of training of the detector for hand postures detection and face detection while wearing glasses. Following such training the classifiers were tested in C++ programming environment. It was found that the classifier or detector could be improved by following a new process of training which was implemented in C++. Retesting with our new method showed remarkable improvements over previous results. Next part of the research under this projects us about development of a reliable method for skin detection. Extensive survey of existing techniques was conducted and a relatively known method for skin detection was successfully implemented. It was used to improve the performances of the face detector. The report concludes by presenting an application that controls the motion of the mouse in Windows XP environment using hand motion, and shows the effectiveness of the implemented techniques. The detector mentioned above was used for developing the application for controlling the mouse pointer was tuned to improve the accuracy of detection. It went through many rounds of the unique improvement process thought up for this project to improve the detection rate of the classifier in order to achieve good accuracy for controlling the mouse pointer in Windows XP environment.