Search

Many approaches to pattern recognition are founded on probability theory, and can be broadly characterized as either generative or discriminative according to whether or not the distribution of the image features. Generative and discriminative models have very different characteristics, as well as complementary strengths and weaknesses. In this paper, we study these models to recognize the patterns of alphabet characters (A-Z) and numbers (0-9). To handle isolated pattern, generative model as Hidden Markov Model (HMM) and discriminative models like Conditional Random Field (CRF), Hidden Conditional Random Field (HCRF) and Latent-Dynamic Conditional Random Field (LDCRF) with different number of window size are applied on extracted pattern features. The gesture recognition rate is improved initially as the window size increase, but degrades as window size increase further. Experimental results show that the LDCRF is the best in terms of results than CRF, HCRF and HMM at window size equal 4. Additionally, our results show that; an overall recognition rates are 91.52%, 95.28%, 96.94% and 98.05% for CRF, HCRF, HMM and LDCRF respectively., {"references":["","M. Elmezain, A. Al-Hamadi and B. Michaelis, Real-Time Capable System for Hand Gesture Recognition Using Hidden Markov Models in Stereo Color Image Sequences, Journal of WSCG, Vol. 16, No. 1, pp. 65-72, 2008.","M. Yang, N. Ahuja and M. Tabb, Extraction of 2D Motion Trajectories and its Application to Hand Gesture Recognition, EEE Trans. on PAMI, Vol. 24, No. 8, pp. 1061-1074, 2002.","Yang, S. Sclaroff and S. Lee, Sign Language Spotting with a Threshold Model Based on Conditional Random Fields, IEEE Trans. on PAMI, Vol. 31(7), pp. 1264-1277, 2009.","C. Sminchisescu, A. Kananujia and D. Metaxas, Conditional Models for Contextual Human Motion Recognition, Journal of CVIU, Vol.104, No. 2, pp. 210-220, 2006.","J. Lafferty, A. McCallum and F. Pereira, Conditional Random Fields: Probabilistic Models for Segmenting and Labeling sequence Data, Conf. on ICML, pp. 282-289, 2001.","L. P. Morency, A. Quattoni and T. Darrell, Latent-Dynamic Discriminative Models for Continuous Gesture Recognition, IIEEE Conf. on CVPR, pp. 1-8, 2007.","X. Deyou, A Network Approach for Hand Gesture Recognition in Virtual Reality Driving Training System of SPG, Conf. on ICPR, pp.519-522, 2006.","M. Elmezain, A. Al-Hamadi, and B. Michaelis, Spatio- Temporal Feature Extraction-Based Hand Gesture Recognition for Isolated American Sign Language and Arabic Numbers, IEEE Symposium on ISPA, pp. 254-259, 2009.","R. R. Lawrence, A Tutorial on Hidden Markov Models and Selected Applications in Speech Recognition, Proceeding of the IEEE, Vol.77(2), pp. 257-286, 1989.\n[10] K. Takahashi, S. Sexi and R. Oka, Spotting Recognition of Human Gestures From Motion Images, In Technical Report IE92-134, pp.9-16, 1992.\n[11] A. Gunawardana, M. Mahajan, A. Acero and J. C .Platt, Hidden Conditional Random Fields for Phone Classification, Proceeding of European Conf. on Speech Communication and technology, pp. 1117-1120, 2005.\n[12] A. Quattoni, S. Wang, L. P. Morency, M. Collins and T. Darrell, Hidden Conditional Random Fields,EEE Trans. on PAMI, Vol. 29, No.10, pp. 1848-1852, 2007.\n[13] L. Rabiner, A Tutorial on Hidden Markov Models and Selected Applications in Speech Recognition, Proceedings of the IEEE, Vol. 77, No. 2, pp. 257-286, 1989.\n[14] X. D. Huang, Y. Ariki, and M. Jack, Hidden Markov Models for Speech Recognition, Edinburgh University Press, 1990.\n[15] S. Goronzy, HRobust Adaptation to Non-Native Accents in Automatic Speech Recognition, Lecture Notes in Computer Sciences, Springer, ISBN-13: 978- 540003250, 2002.\n[16] M. Elmezain, A. Al-Hamadi, and B. Michaelis, Discriminative Models-Based Hand Gesture Recognition, International Conference on Machine Vision, pp. 123-127, 2009.\n[17] A. McCallum, D. Freitag, and F. Pereira, Maximum Entropy Markov Models for Information Extraction and Segmentation,International Conference on Machine Learning, pp. 591-598, 2000.\n[18] J. Lafferty, A. McCallum, and F. Pereira, Conditional Random Fields: Probabilistic Models for Segmenting and Labeling sequence Data, International Conference on ICML, pp. 282-289, 2001.\n[19] A. McCallum, Efficiently Inducing Features of Conditional Random Fields, Conf. on Uncertainty in AI, 2003.\n[20] L. P. Morency, A. Quattoni, C. M. Christoudias, and S. Wang, Hidden-state Conditional Random Field Library, EVersion 1.3c, http://pt.sourceforge.jp/projects/sfnet hcrf/,” 2008."]}

Posture and Gesture Recognition for Human-Computer Interaction

Automatic gesture spotting and recognition is a challenging task for locating the start and end points that correspond to a gesture of interest in Human-Computer Interaction. This paper proposes a novel gesture spotting system that is suitable for real-time implementation. The system executes gesture segmentation and recognition simultaneously without any time delay based on Hidden Markov Models. In the segmentation module, the hand of the user is tracked using mean-shift algorithm, which is a non-parametric density estimator that optimizes the smooth similarity function to find the direction of hand gesture path. In order to spot key gesture accurately, a sophisticated method for designing a non-gesture model is proposed, which is constructed by collecting the states of all gesture models in the system. The non-gesture model is a weak model compared to all trained gesture models. Therefore, it provides a good confirmation for rejecting the non-gesture pattern. To reduce the states of the non-gesture model, similar probability distributions states are merged based on relative entropy measure. Experimental results show that the proposed system can automatically recognize isolated gestures with 97.78% and key gestures with 93.31% reliability for Arabic numbers from 0 to 9.

A Kalman filter is a recursive estimator and has widely been used for tracking objects. However, unsatisfying tracking of moving objects is observed under complex situations (i.e. inter-object merge and split) which are challenging for classical Kalman filter. This paper describes a multi-hypothesis framework based on multiple features for tracking the moving objects under complex situations using Kalman Tracker. In this framework, a hypothesis (i.e. merge, split, new) is generated on the basis of contextual association probability which identifies the status of the moving objects in the respective occurrences. The association among the moving objects is computed by multi-featured similarity criteria which include spatial size, color and trajectory. Color similarity probability is computed by the correlation-weighted histogram intersection (CWHI). The similarity probabilities of the size and the trajectory are computed and combined with the fused color correlation. The accumulated association probability results in online hypothesis generation. This hypothesis assists Kalman tracker when complex situations appear in real-time tracking (i.e. traffic surveillance, pedestrian tracking). Our algorithm achieves robust tracking with 97.3% accuracy, and 0.07% covariance error in different real-time scenarios.