Eye Center Localization using cascaded corner detection and geometrical measurements algorithm

Abstract

The eye center localization can be achieved using the corner detection algorithm, which is an algorithm intends to find the corner points of a face and use of corner detection algorithm is to mark a point, where the line should be drawn on parts of a face. The corner detection algorithm used in this research work performs the task of finding the corners of a face such as eyes, nose, and mouth, but the paper mainly focuses on the corners of an eye using an eye detection algorithm, as the paper intends to find the center of an eye. The eye detection algorithm is required to consider only the points of our interest. The corner points obtained using corner detection algorithm is given to eye detection algorithm, which considers only the corner points that are found near eyes, and these points are used as a reference to draw a rectangle using geometrical measurement algorithm. The geometrical measurement is another method employed in this research work to draw a rectangle around the corner points of two eyes. The output of geometrical measurement algorithm is an exact center of two eyes. All three algorithms have been linked to one another. The output of corner detection algorithm is given to eye detection algorithm, which in turn gives its output to geometrical measurements algorithm. Since there is a cascading of output from one algorithm to another, the method is collectively called as Eye Center Localization with cascaded corner detection, and geometrical measurements algorithm. The accuracy achieved during the process of localizing the center of an eye is 99.64%, which is better than other approaches to the best of our knowledge.