基于肤色的人脸检测算法研究

文章提出了一种基于肤色的彩色图像人脸检测算法。该算法会对输入图像进行尺寸归一化及光线补偿,综合利用2种色彩空间对预处理后的图像进行二值化操作。在肤色分割的基础上去除噪声,并结合多种有用信息,对肤色区域进行了判定,最终识别出人脸区域。它的主要优点有对人脸的姿态不敏感、检测速度快等。在多人脸检测方面,该算法运用区域分割的方法对多个人脸进行有效检测。实验结果表明,该算法是可行有效的。

Study of face detection algorithm based on complexion

A complexion based face detection algorithm for chromatic images is proposed in this paper.Firstly,the input image is processed by the dimension unification operation and light compensation.Secondly,a binary-value transformation operation is executed upon the preprocessed image by utilizing two color spaces.Then the noises are eliminated based on the complexion partitioning,and the complexion areas are judged according to various useful factors.Finally,the human face areas are recognized.This method has many advantages such as the insensitivity to the face posture and the rapid inspection speed etc.In the multi-face detection field,this algorithm uses the area partitioning method to effectively detect the multi-face.The experimental results demonstrate that the proposed algorithm is feasible and effective.