Commenced in January 2007
Frequency: Monthly
Edition: International
Paper Count: 30184
Posture Recognition using Combined Statistical and Geometrical Feature Vectors based on SVM

Authors: Omer Rashid, Ayoub Al-Hamadi, Axel Panning, Bernd Michaelis

Abstract:

It is hard to percept the interaction process with machines when visual information is not available. In this paper, we have addressed this issue to provide interaction through visual techniques. Posture recognition is done for American Sign Language to recognize static alphabets and numbers. 3D information is exploited to obtain segmentation of hands and face using normal Gaussian distribution and depth information. Features for posture recognition are computed using statistical and geometrical properties which are translation, rotation and scale invariant. Hu-Moment as statistical features and; circularity and rectangularity as geometrical features are incorporated to build the feature vectors. These feature vectors are used to train SVM for classification that recognizes static alphabets and numbers. For the alphabets, curvature analysis is carried out to reduce the misclassifications. The experimental results show that proposed system recognizes posture symbols by achieving recognition rate of 98.65% and 98.6% for ASL alphabets and numbers respectively.

Keywords: Feature Extraction, Posture Recognition, Pattern Recognition, Application.

Digital Object Identifier (DOI): doi.org/10.5281/zenodo.1331595

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1177

References:


[1] R. Bowden, A. Zisserman, T. Kadir and M. Brady, Vision Based Interpretation of Natural Sign Languages, in Proc. 3rd International Conference on Computer Vision Systems, 2003, pp. 391-401.
[2] M. Hussain, Automatic Recognition of Sign Language Gestures, Master Thesis, Jordan University of Science and Technology, 1999.
[3] M. Handouyahia, D. Ziou and S. Wang, Sign Language Recognition Using Moment-Based Size Functions, in Proc. International Conference of Vision Interface, 1999, pp. 210-216 00.20.40.60.810 50 100 150 200 250 Circularity Rectangularity Frames Values 32 54 24 100.20.40.60.810 50 100 150 200 2500 1 2 3 4 5 6FramesProbabilityFrame 43Frame 9 Frame 70 Frame 120Frame 140 Frame 167 Frame 183Frame 170 Frame 230Frame 80
[4] S. Malassiotis and M. Strintzis, Real-time Hand Posture Recognition using Range Data, in Image and Vision Computing, Vol. 26, No. 7 pp. 1027-1037, 2008.
[5] A. Licsar and T. Sziranyi, Supervised Training Based Hand Gesture Recognition System, in Proc. International Conference on Pattern Recognition, 2002, pp. 999-1002.
[6] W. Freeman and M. Roth, Orientation histograms for hand gesture recognition, in Proc. International Workshop on Automatic Face and Gesture Recognition, 1994, pp. 296-301.
[7] M. Hu. Visual Pattern Recognition by Moment Invariants, in IRE Transaction on Information Theory, Vol. 8, No. 2, pp. 179-187, 1962.
[8] S. Maitra, Moment Invariants, in Proc. of the IEEE, Vol. 67, pp. 697-699, 1979.
[9] J. Flusser and T. Suk, Pattern Recognition by Affine Moment Invariants, in Journal of Pattern Recognition, Vol. 26, No. 1, pp. 167-174, 1993.
[10] J. Davis and G. Bradski, Real-time Motion Template Gradients using Intel CVLib, in Proc. of IEEE ICCV Workshop on Framerate Vision, 1999.
[11] N. Cristianini and J. Taylor, An Introduction to Support Vector Machines and other kernel based learning methods, Cambridge University Press, 2001.
[12] C.J. Lin and R. Weng, Simple Probabilistic Predictions for Support Vector Regression, in Technical Report, Department of Computer Science, National Taiwan University, 2004.
[13] Point Gray Research. (2008, Nov 8). Bumblebee2 stereo vision camera. Available: www.ptgrey.com/products/Point_Grey_stereo_catalog.pdf