Study On Human Eye Segmentation And Pupil Location

Eyes,the windows of soul,can convey the inner word of mankind.The study of eye is the key to understand the human emotions,behavior,visual mechanisms,and activities of heart.The application of pupil detection and positioning is most widely used in many applications of human eye research.It involves artificial intelligence,cognitive psychology,virtual reality and many other disciplines.What is more,it has a high application prospects in the industrial,medical testing,disabled access,virtual reality and so on.Fast and accurate positioning of the pupil is still a hot topic of research topics.At present,some methods of eye segmentation and pupil positioning can achieve better results under ideal conditions,such as the collection of images at a close proximity of special equipment,or the location of the human eye.Compared with images obtained in ideal conditions,the image obtained by ordinary equipment is not better and the boundaries of pupil and iris are not so clear.The positioning of the pupil is not affected by light,occlusion,pupil position,and eye condition under ideal conditions.This method does not apply to large-scale promotion due to special equipment price security,and fixed eye position will limit human action.External factors will have a great impact on the pupil positioning in the actual situation,such as light distribution,eyelid eyelash occlusion,spot and so on.While the state of the eye itself will also affect the positioning of the pupil,such as the eye to see a different direction,the pupil in the eyes of different locations,different shapes,different degree of opening,eyelash occlusion and other effects.The positioning of the pupil will be affected due to the different state of the eye,the image is not clear due to the ordinary collection of equipment,and these are the problems.This paper presents a new approach to positioning pupils from the face.The main idea of the method is follow.We use the anthropometric method to find the area that contained eye and eyebrow in face detection.Multiple thresholds are acquired in the grayscale histogram and the binary images of every threshold are obtained too.The valid regions are fused in multiple binary images and then get the candidate area.The information entropy is used to determine the eye area to separate of eyes and eyebrows.According to the characteristics of the eye,the appropriate threshold is selected and the pupil is divided by threshold.We use the projector to find the pupil center.We divide the pupil edge equally into six points with the center,and we randomly selecte a point set at each aliquot,and we get the pupil center with fitting these points through the ellipse.First of all,this article on the eye detection and pupil positioning of the topic background and the significance of the study were introduced in detail.We summarize the methods of recent years and compare their advantages and disadvantages in two aspects that the eye of the rough positioning and precise positioning.Then,this paper proposes a method of segmenting and positioning the eye based on the method of regional fusion.The eyes are segmented with the information entropy of different regional fusions.The experimental results show that the eyes in different directions in the method can effectively segment and locate the eye.Finally,this paper introduces the basic concepts and mathematical principles of ellipse fitting,and uses ellipse fitting to locate the pupil.We select the appropriate threshold according to the structure of the eye,and split the pupil.After splitting the pupil,the pupil is roughly positioned by projection integration.We select the left and right edges of the center to ellipse fit the pupil.The experimental results show that the method can fit the complete pupil contour and locate the pupil.