A novel real-time eye detection method using edge detection and Euclidean distance

This work presents a novel approach to the recognition of eyes and mouths using Euclidean distance and edge detection. By face detection methods, the face region is detected. By merging the outcomes of the Sobel and Prewitt masks, horizontal edges can be detected. Some aesthetic operations like dilation and erosion are used to remove smaller edges and join closer edges together. The screen area is divided into the top right, top left, and bottom. The right eye, left eye and mouth are the largest binary objects of each region respectively. The distance between the right eye center and mouth center is compared with the distance between the left eye center and mouth center as a verification step. The accuracy of the method based on the Aberdeen database is 92%, which is the highest accuracy for the face detection method.