基于Adaboost人脸检测融合五官特征的性别识别

Gender recognition based on Adaboost face detection integrated with facial features

为了有效地利用人脸的五官特征以提高人脸检测与性别识别的性能,提出一种自然场景下的人脸性别识别算法.首先通过级联低阈值的人脸检测器和肤色、五官检测器来对人脸进行检测,并利用五官特征对人脸进行矫正、标定;然后对人脸图像进行LBP特征提取,并通过加入权重集来增大特征中五官特征的权重,提高识别率.检测结果表明:该算法在自行建立的OWN数据库中有较好的人脸检测率;同时在公开的FERET及LFW数据库上均取得了较高的性别识别率,且对自然场景中的光照、人脸表情和位置的变化具有较高的鲁棒性.

In order to effectively utilize the facial features to improve the performance of the face detection and gender identification,a novel algorithm for face gender recognition in the natural scene was proposed.Firstly,a low threshold face detector,a skin color filter and facial features detector were cascaded for the face detection.Afterwards facial features were used to correct and calibrate the face image.Then the LBP(local binary pattern)features were extracted from the calibrated face image, which was weighted by the facial features to improve the recognition rate.The results show that the algorithm not only has a better face detection rate on OWN database established by ourselves,but also has achieved a high level of gender recognition rate on public FERET(face recognition technology) and LFW(lableled faces in the wild home)database,with high robustness to illumination,facial expression and location in the natural scene.