提出一种在检测到人脸区域的前提下,对人眼进行准确定位的算法。该算法首先计算人脸图像的相似度信息,并对得到的相似度图像进行二值化,从而标定人脸区域。然后对得到的人脸区域进行图像块分割,通过复杂度计算得到三个复杂度最大的区域块,计算其居中度,从而实现对人眼的准确定位。
An eye location algorithm was developed on the premise of that the face region has been detected. First the algorithm calculates the similarity information and gets the binary image to locate the face region. Then it segments the face region to several blocks and selects the three most complexity blocks, calculates their central degree to complete the eye location.