Abstract
In healthcare systems, video surveillance, interactive games and many other applications, human action recognition and recording are very important. This paper presents a comprehensive 3D model based action recording system using computer visions. First, a computer captures human motion videos with a network camera and conducts further detection and tracking of the video resources, then a 3D model is created based on the recorded data results. The action recording system includes background capturing, characters capturing, action tracking, 3D modeling and OGRE controlling. OpenCV is used for background and characters capturing where a background image difference algorithm is used to analyze a moving target and extract different elements. For the action tracking, the Camshift (Continuously Adaptive Mean shift Algorithm) tracking algorithm is used to realize continuous tracking and recognition of moving objects and ensure good performance of the action recorder. In our implementation, 3Dmax is used to build a 3D model and skeletal animations, where Ogremax is used to export models, and then to import the skeletal animations into a testing environment. The evaluations show that our motion recognition and recording system has good performance in one aspect, and can obtain accurate result on the other aspect.
Keywords
References
Mao, J., Zou, X., Zhu, J.: Improved Plane Template Two-step Camera Calibration. China Journal of Image and Graphics 9(7), 846–852 (2004)
Zhang, C., Su, Y.: Digital Image Compression, pp. 20–23. Tsinghua University Press (2006)
Park, J.S.: Interactive 3D reconstruction from multiple images: a primitive-based approach. Pattern Recognition Letters 26(16), 2558–2571 (2005)
Intel Corporation. Open Source Computer Vision Library Reference Manual (December 2001)
de la Torre Frade, F., Marroquín, E.M., Pérez, M.E.S., Moreno, J.A.M.: Moving object detection and tracking system: a real-time implementation (1997)
Nascimento, J.C., Marques, J.S.: Performance evaluation of object detection algorithms for video surveillance. IEEE Trans. Multimedia, 761–774 (2006)
Wu, F., Li, H., Hu, Z.: A New Camera Self-calibration Method Based on Active Vision System. Chinese Journal of Computers 23(11), 1130–1139 (2000)
Feng, L., Wang, R.: Application of CCD Sensors in the Image Detection System of Human Motion. Instrument Technique and Sensor 5, 33–35 (2000)
Wang, T., Feng, W., Luo, D.: Application of Wireless Sensor Network for Detection of Human Body Movements. Instrumentation and Measurment 28(8), 83–86 (2009)
Pi, W., Liu, H., Cha, H.: Motion Detection for Human Bodies Basing Adaptive Background Subtraction by Using an Omnidirectional Camera. Journal of Peking University 40(3), 458–464 (2004)
Cao, D., Zou, W., Wu, Y.: Motion human detection based on difference background image. Opto-Electronic Engineering 34(6), 107–111 (2007)
Liu, Y., Ai, H., Xu, G.: Moving object detection and tracking based on background subtraction. Information and Control 31(8), 315–319 (2002)
Wei, J., Liu, Z., Zhang, H.: Moving objects detection algorithm under complex background environment. Microcomputer & Its Applications 2, 49–51 (2010)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2013 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Li, Y., Rao, Y., Duan, Y., Zhang, W. (2013). A 3D Model Based Action Recorder Using Computer Vision. In: Ghose, A., et al. Service-Oriented Computing - ICSOC 2012 Workshops. ICSOC 2012. Lecture Notes in Computer Science, vol 7759. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-37804-1_29
Download citation
DOI: https://doi.org/10.1007/978-3-642-37804-1_29
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-37803-4
Online ISBN: 978-3-642-37804-1
eBook Packages: Computer ScienceComputer Science (R0)