Abstract
This paper asserts a method of controlling the cursor based on the Fisher linear discriminant (FLD) analysis to recognize six classifications of human face movements, which are face movements upwards, downwards, leftwards, rightwards, blinking of the right eye and blinking of the left eye. These classifications represent cursor movements upward, downwards, leftwards, rightwards, and the right or left button on the mouse respectively. This method can be separated into two areas: face detection and gaze recognition. Face detection is to convert the RGB color space into YCbCr in order to segment skin tone areas, where the human face area is located by using the connect component labeling method. Gaze recognition is accomplished by building a gaze recognition training model parameter through FLD algorithm. Then, by using the Euclidean distance as the rule of decision, match the detected facial image to the parameters of this model to find the shortest Euclidean distance and its corresponding classification to control cursor movements.
Publisher
Trans Tech Publications, Ltd.
Reference5 articles.
1. C. N. Kuo: Development and Application of A Multi-functional PC Camera with Eye-tracking and Head Controlled System, Master thesis of Department of Automatic Control Engineering, Feng Chia University, Taiwan (2001).
2. P. Ballard, and G. C. Stockman: Computer operation via face orientation, Proc. IEEE conf. on Pattern Recognition, Vol. 1 (1992), pp.407-410.
3. J. Kim, K. R. Park, J. J. Lee and S. R. LeClair: Intelligent process control via gaze detection technology, Engineering Applications of Artificial Intelligence, Vol. 13, no. 5 (2000), pp.577-587.
4. T. Morris, and V. Chauhan: Facial feature tracking for cursor control, Journal of Network and Computer Applications, Vol. 29, no. 1 (2006), pp.62-80.
5. C. Xiang, X. A. Fan, and T. H. Lee: Face recognition using recursive Fisher linear discriminant with Gabor wavelet coding, Proc. IEEE conf. on Image Processing, Vol. 1 (2004), pp.79-82.