A rubost and fast eyes locating algorithm

The performance of existed eyes localization algorithm degrades obviously when influenced by the variation of background, the non-uniformity of illumination, the change of facial posture and eyebrows and so on. To address these problems, a fast and efficient eye location algorithm is proposed in this paper. Firstly, based on cascade Adaboost face detection algorithm, the facial areas surrounding eyes are obtained using the analysis of gradient complexity, and then Gabor filter, integral projection and connected region enhancement algorithm are employed to locate the eyes accurately. Experiments show that the method is adaptive to the changes of head posture and facial expression, with detection accuracy over 99%.