比较Opencv自带的eye检测器

CascadeClassifier *lbpcascade_frontalface = new CascadeClassifier("./xml/lbpcascade_frontalface.xml");
CascadeClassifier *haarcascade_eye = new CascadeClassifier("./xml/haarcascade_eye.xml");
CascadeClassifier *haarcascade_eye_tree_eyeglasses = new CascadeClassifier("./xml/haarcascade_eye_tree_eyeglasses.xml");
CascadeClassifier *haarcascade_mcs_eyepair_big = new CascadeClassifier("./xml/haarcascade_mcs_eyepair_big.xml");
CascadeClassifier *haarcascade_mcs_eyepair_small = new CascadeClassifier("./xml/haarcascade_mcs_eyepair_small.xml");
CascadeClassifier *haarcascade_lefteye_2splits = new CascadeClassifier("./xml/haarcascade_lefteye_2splits.xml");
CascadeClassifier *haarcascade_righteye_2splits = new CascadeClassifier("./xml/haarcascade_righteye_2splits.xml");
CascadeClassifier *haarcascade_mcs_lefteye = new CascadeClassifier("./xml/haarcascade_mcs_lefteye.xml");
CascadeClassifier *haarcascade_mcs_righteye = new CascadeClassifier("./xml/haarcascade_mcs_righteye.xml");
采用opencv自带的LBP人脸检测器进行检测，将检测结果的上半部分输入眼睛检测器，检测结果一次为

eye\eye_tree_eyeglass\

eye_tree_eyeglass检测结果较好，但是出现漏检

mcs_eyepair_big\mcs_eyepair_small\

eyepair是检测眼镜使用的，由此检测可以看出，检测结果未必准确。

lefteye_2split\mcs_lefteye\

righteye_2split\mcs_righteye\

相对于split，mcs的检测结果要好很多。