•  
  •  
 

Turkish Journal of Electrical Engineering and Computer Sciences

DOI

10.3906/elk-1409-10

Abstract

In this paper, a fast and accurate algorithm is proposed to recognize open and closed eye states. In the proposed algorithm, first a hierarchical preprocessing stage is used to detect eye areas. This stage employs Haar features to detect face area, color, and intensity mappings to extract eye candidate areas, and some simple geometrical relations for a final decision of the eye regions. In the second stage of the algorithm for detecting eye state, a new proposed descriptor based on a histogram of local maximum vertical derivative patterns in eye areas is extracted and applied to a support vector machine classifier. The proposed descriptor, while having low computational complexity, is defined well enough to describe eye features and hence can distinguish well between open and closed eyes. Experimental results from test images show that the proposed algorithm can correctly detect the eye state by the rate of 98.2%, which is higher than other similar algorithms.

Keywords

Eye detection, eye state recognition, local maximum vertical derivative pattern, support vector machine classifier

First Page

5124

Last Page

5134

Share

COinS