【摘要】 提出了一种快速的、鲁棒的人脸定位及跟踪研究方法 ,定义了一种新的运动能量表示方法 ,利用该方法可以很快地检测出图像中的运动区域。在检测到的运动区域内 ,利用人的头部及面部结构知识确定出人脸区域 ,再将检测到的人脸区域作为模板 ,利用遗传算法在新的图像序列中动态地进行模板匹配 ,以确定人脸在新的图像序列中的位置。实验结果表明 ,该方法不但检测速度快 ,而且对光照、人脸姿态、表情及其他噪声具有较强的鲁棒性更多还原

【Abstract】 A fast and robust approach method for making the detection, localization and tracking of a persona face in image sequences was presented. A new representation of motion energy was defined. This method can be used to extract the moving region in image in real time. Then, the face area of a persona can be determined by the knowledges about the hair and head of the persona. For face tracking, a new genetic algorithm based dynamic template matching method was applied to search the new position of the face of a persona in each new video sequence. Experimental results show that the proposed method is fast and robust in description of illumination, face poses, facial expressions and image distraction, such as facial occlusion by hands.更多还原