Hand Tracking is a great application that gets most of its attention in recent times. Hand tracking can be used in many applications like gesture interactions, robotic arms and etc. This project proposes a real-time hand tracking algorithm using skin tone segmentation. We use the viola jones algorithm to detect face and skin color is detected from the cropped image. Image segmentation is performed using skin color on the whole frame and finally, hand is segmented from the final background subtracted image. This approach is relatively insensitive to the background, achieving robust tracking performance in real-time.
- Principal Component Analysis
- DCT and shape features
Drawbacks of Exisitng System
- High Computational load and poor discriminatory power.
- slow training for a large feature set.
- Less accuracy in classification
- Feature matching
- The segmentation algorithm Proves to be simple and effective
- The greyscale Co-occurrence matrix performed well
- Better texture and edge representation
- Segmentation provides better clustering efficiency
- 4 GB of RAM
- 500 GB of Hard disk
- MATLAB 2014a
 J. Han and M. Kamber, Data Mining Concepts and Techniques, United States of America, 2001.
 S.Cheng, C. Hsu, and J. Li, “Combined Hand Gesture-Speech Model for Human Action Recognition,” in Sensors, vol 13, 2013.
 P. Fankhauser, M. Bloesch, and D.Rodriguez, “Kinect v2 for Mobile Robot Navigation: Evaluation and Modeling,” in Advanced Robotics (ICAR), International Conference on. IEEE, 2015., 2015.
 V. Vapnik, The Nature of Statistical Learning Theory, New York: Springer-Verlag, 1995.
 J.C.Platt, N. Cristianini, J. Shawe-Taylor, “Large margin dags for multiclass classification.,” in Advances in Neural Information