一种快速人眼精确定位算法

An algorithm for rapid eye detection

文中提出了一种基于最小外接矩形算法的人眼精确定位方法,该算法将顶点链码与离散格林理论相结合用于快速提取目标图像的最小外接矩形、求出目标图像的面积、形心和主轴。本文首先采用肤色分割与灰度投影相结合的方法快速定位到人脸,然后根据人脸的水平灰度投影曲线确定出人眼的大致位置,最后利用本文提出的算法实现了人眼的精确定位。实验结果表明该算法能快速、有效地进行人眼精确定位。

An algorithm based on the Minimum Enclosing Rectangle （MER） for detection and recognition of human eyes is proposed in this paper.According to this algorithm,it can calculate the area,centroid,and principal axis of the image quickly.At first,this paper adopt the method of skin color segmentation combined with gray-scale projection to locate the face,and then according to horizontal gray-level projection curve of the face,the general location of the eyes can be determined quickly.Finally,the precise localization of human eyes can be realized by using the proposed algorithm.Experiments show that the exact positions of human eyes are located rapidly by using this algorithm.