基于肤色检测和人眼定位的人脸检测方法

提出了一种基于肤色检测和人眼定位的人脸检测方法。使用基于“基准白色”的色彩平衡方法归一化彩色图像,将图像在HSV空间进行肤色分割,确定候选人脸,采用形态学滤波器对其降噪。在获得虹膜位置的基础上,使用Susan算子定位两个眼角点,从而实现眼睛的精确定位。实验证明提出的方法能够很好的检测人脸、定位人眼,尤其对存在人脸旋转和光照异常的人脸图像有很高的精确度和鲁棒性.

Face Detection Based on Skin Color Detection and Eyes Localization

Presenting a method of face detection based on skin color detection and eyes localization, this paper introduces a method of color balancing to normalize the color image called "reference white". It extracts skin color in the HSV color space and detects the candidate human face. It uses the morphologic filter to denoise. After iris localization it introduces Susan operator to localize the corners of eyes, then implements the localization of eyes accurately. The experiments demonstrate that the method in this paper can implement face detection and eyes localization commendably, especially in the image of face circumrotation and illumination exceptional.