基于人脸特征和Adaboost算法的人脸定位

Face Detection Based on Adaboost Algorithm and Facial Features

人脸检测与定位,是人脸分析的基础,应用在众多领域,一种快速准确高效的人脸定位算法具有重要的意义。文中提出了一种利用人脸的特征与改进的Adaboost算法相结合的人脸定位算法。首先利用人脸的肤色特征,寻找出人脸可能存在的区域。再利用人形态特征筛选出人脸的候选块。针对人脸定位中普遍存在的人脸偏转,文中利用人脸的对称性予以调整,最后利用改进的Adaboost算法在候选块中快速定位人脸。实验表明该算法多姿态,复杂环境下的人脸具有良好的检测效率和较高的准确性及鲁棒性。

Face detection and location is the basis of face analysis and applications in many field. So one fast accurate and efficient face detection algorithm is of great significance. It proposes a human face detection by making use of facial features and Adaboost algorithm. First,search the regions where might exist faces by using skin color information. Then make use of some morphological to search the pos- sible face blocks. Finally, Adaboost algorithm is used for fast decision on the candidate. Experiment result shows that the proposed algo- rithm is fast,and there is a better accuracy and robusmess on copying with multi-profile,comPlex backgrounds face.