【摘要】 随着3D显示技术的发展,在显示技术中需要对人脸进行精确检测以及对五官进行精确定位。提出了将亮瞳效应、运动帧间差分投影和多模板ASM匹配集于一体的人脸检测方法。首先利用亮瞳效应找出可能包含的人脸区域;然后利用差影法对可能包含的区域进行进一步筛选缩小范围;最后利用多模板ASM方法进行精确定位。此方法发挥各种检测算法的优势,准确率较高,对姿态、表情、背景等变化情况下人脸的检测具有较好的鲁棒性。更多还原

【Abstract】 With the development of 3D display technology,technologies are required to perform accurate detection of human faces and accurate positioning of the features.A human face detection method based on bright pupil effect,frame difference and ASM matching is proposed.Firstly,the region which may contain faces is predicted by bright pupil effect.Then difference image is used to minish the searching area of faces.Lastly,ASM templates are used to match and verify.Experiment results proved that our algorithm can overcome the variability of lighting condition,pose,facial expression and background by taking the advantages of various detection algorithms.更多还原