The use of hand gestures opens a wide range of application for human computer interaction. The paper makes use of haar classifiers and camShift algorithm to track the movement of hand. Parallelism is introduced at every step by segmenting the data from camshaft into an NxN grid. Every block of the grid now represents a lead point which is calculated from mean of all the points belonging to the particular grid. Now we have only N2 points to recognize the curve that was performed by the user in his action. Finally the fit that was found is compared to pre-defined curve fit data to find out the curve using Mahalanobis equation. Parallelism used in reducing the number of points to be fitted allows the recognition to be faster.
Agarwal, Bhavish Sushiel; Desai, Jyoti R; and Saha, Snehanshu
"A Fast Haar Classifier based Gesture Recognition using camShift algorithm and Curve Fitting Method,"
International Journal of Computer Science and Informatics: Vol. 1
, Article 14.
Available at: https://www.interscience.in/ijcsi/vol1/iss4/14