基于肤色检测和Adaboost联级算法的人脸定位研究

Study on Face Location Based on Skin Color Detection and Adaboost Cascade Algorithm

提出了一种基于人脸肤色信息结合Adaboost联级检测的人脸检测算法,对彩色图像中的人脸进行定位。首先,该方法在YCb Cr彩色空间下,利用肤色检测粗定位人脸,过滤掉一些非肤色区域,然后对肤色检测处理后的图像进行二值化,并以此作为模板,对原图像进行掩膜处理,再用Adaboost联级算法对掩膜后的图像进行检测。检测结果证明,所用的方法缩小了Adaboost联级算法的检测区域,提高了检测速度,并且减少了单一的Adaboost联级算法的误检率,可以在一定程度上避免Adaboost联级算法对大量非人脸样本的依赖性,从而减少训练时间。

This paper put forward a kind of face detection method based on the skin color information and Adaboost cascade detection algorithm, carried out the localization of human face in color images. First, the method used skin color detection to coarsely position the face in YCbCr color space, filtered out some non-skin regions, and then carried out the binarization of skin color detection image, and masked the original image, then used the Adaboost cascade algorithm to detect the masked image. Detection results prove that this method reduce the detection region of the Adaboost cascade algorithm, and accelerate the speed of detection. At the same time, it can reduce the single Adaboost cascade algorithm＇ s false detection rate, avoid the dependency of Adaboost cascade algorithm on the vast of non-face samples in a certain extent, and then reduce the training time.