人脸姿态分析与仿真

【作者】 齐金山；

【导师】 孙兴华；

【作者基本信息】 南京理工大学 ， 计算机应用技术， 2008， 硕士

【摘要】 驾驶员的工作状念是决定行车安全的重要因素之一,由于驾驶员工作状态不良而引发的交通事故,主要原因在于驾驶员的疲劳和精神疏忽,而脸部的姿态运动是关键性因素之一。本文的研究内容是使用一个普通的摄像头对人的脸部姿态进行实时的分析,并通过OpenGL三维建模对人脸姿态分析算法进行验证,以此来判断驾驶员当前的驾驶状态是否存在不安全因素。首先我们检测出镜头内的人脸区域,通过边缘检测及模板匹配算法检测出人眼和嘴角区域,以此为基准计算脸的偏转方向。接下来,通过OpenGL建立三维的人脸模型,把人脸姿念的运动参数实时的表现出来。最后,我们对不安全的状态予以提示。经验证,在实验室环境下,本系统能较准确的检测出人脸以及各特征点,并应用仿射变换原理对人脸的姿念进行了估计以及对人脸姿态进行实时的仿真。更多还原

【Abstract】 The work state of drivers is one of the most important factors of safety during driving. There are two important reasons which result traffic accident. They are tiredness and carelessness of drivers, which reflected by face pose where the shoe pinches. This paper main presents the real time analysis of face pose using an ordinary camera and tests the results by 3D modeling using opengl, by means of which judging whether driver is at safety state.First, we recognize the whole face in camera lens. Then eyes and mouth are recognized by edge detection and template match which are looked upon as benchmark counting how the face rotated. The 3-D model shows the real time parameters of face pose. At last we warn of the insecurity states. The system is efficient at experimentation situation. By experiments in our lab’s platform, the system proposed in this paper can detect out each face feature points exactly, and apply the principle of affine transformation to estimate the stance of the faces and realize the real-time simulation of the faces.更多还原