基于肤色及AdaBoost算法的自动调焦人脸检测

Human-face Autofocus Detection Based on Complexion and AdaBoost Algorithm

提出一种结合肤色检测和AdaBoost算法的自动调焦实时人脸检测方法。采用肤色信息定位皮肤区域,针对AdaBoost算法对待检人脸尺寸小于训练样本人脸尺寸时检测率低的缺陷,并采用自动调焦算法调整图像大小后再利用Ad-aBoost算法进行人脸检测。实验结果表明,结合肤色信息和自动调焦后的AdaBoost算法性能较传统AdaBoost算法有明显提高,同时对小人脸图像也有更好的检测效果。

An autofocus and human face real time detection method combining complexion detection with AdaBoost algo- rithm is proposed. The complexion information is adopted to locate the area of skin, and then the autofocus algorithm is uti- lized to adjust the image size before using AdaBoost algorithm to detect human-face in view of the low detection rate when the size of human-faces is smaller than that of training samples. The experiment results show that in comparison with the per- formance of traditional AdaBoost algorithm, the performance of AdaBoost algorithm combining skin color information with autofocus method is significantly improved. Tthe better effect of small face image detection is also acquired.