With the development of computer vision technology, non-invasive eye positioningand tracking algorithm is taking a critical step for the commercial applications of HCIand widely applying the applications in the services of home, office and social. In orderto improve the subjective experience of users, computer vision applications put forwardhigher requirements to the accuracy and real-time of the human eye positioning.In this paper, we proposed a new eye location method based on a step of facedetection, as well as using the technology of corner detection and integration projection.Firstly, we use Haar features and Adaboost algorithm, followed by a image croppingand preprocessing processing method, to extract the regions of interest from a faceimage. Then an improved method based on the corner detection is proposed to get therough area of eyes. The accurater location of eyes is judged and adjusted by locating thetwo eyes separately with integration projection at last. In addition, this paper furtheruses the CamShift algorithm and the adaptive binaryzation threshold processing torealize a human eye positioning in the video environment. Compared with the latesttechnologies based on the appearance, feature and the IR illumination, the proposedalgorithm can overcome the interference from the eyes closed, light and slight rotationof the head and track eyes robustly with a higher accuracy. |