一种基于显著性和部件模型的无约束条件人脸检测方法

A Method Using Saliency and Part-based Model for Automatic Face Detection in Unconstrained Conditions

针对无约束条件下的人脸检测进行研究,提出了一种基于显著性度量和部件模型的人脸检测方法。在部件模型为基础的检测方法中引入显著性理论,融合正面、左侧面、右侧面的三种平面外旋转姿态下的信息,生成完整有效的人脸显著图,并用于人脸检测。实验结果表明本文生成的人脸显著图能够更有针对性描述人脸区域,并且本文人脸检测方法相较Viola Jones人脸检测方法和基于部件模型的人脸检测方法有更高的检测率,运算速度相较原基于部件模型的人脸检测方法大有提高。

Face detection on unconstrained conditions is studied, and a method using saliency and calculation model of face saliency is proposed based on part-based model and the theory of saliency. Face saliency maps under three perspectives out of plane for each test image are generated, then a complete and effective face saliency map is received to achieve the purpose of face detection. The results show that Through the constructed face saliency map the facial region can be describe accurately, and our face detection method honors the better performance compared to the face detector of Viola Jones and method based on part-based model, and our improved method performs higher operation speed compared to the original part-based model.