The Application of Support Vector Machines in the Automatic Eye Position Algorithm

Support vector machine (SVM) was a new and outstanding machine learning as an efficient machine learning tool in dealing with small samples. In this paper, an new automatic eye position algorithm based on SVM is introduced, which is fast and accurate and the eyeball´s center position velocity is only 1 second. The position accuracy is up to 95 percent and average position error is about 3 pixels. Compare to existing eye localization algorithm, the algorithm mentioned in this paper is simple and easy to implement for position. The experimental results show that this new method is satisfying in accuracy of the automatic eye position, and using this algorithm, the position velocity is faster and position accuracy is higher than other eye position algorithm.