摘要
多姿态人脸识别是当前人脸识别中的难点 ,识别率普遍不是很高。本文提出了一种利用正面、侧面照片建立起三维虚拟人脸模型 ,然后对待识别照片进行角度估计 ,把模型库中的每一个 3D模型在该角度附近进行多次投影 ,搜索出与待识别照片相似度最高的照片。这种方法可以有效的提高人脸识别率 。
Posevaried Face Recognition is a nut of face recognition and has a low ratio of recognition at present A way is presented by building the 3D virtual person model using front and side face photo and then making the projection of the each 3D model near the rotating angle estimated by the unknown photo for many times and finding out the similar photo Recognition ratio can be raised effectively and system′s robust is increased also by this method
出处
《现代电子技术》
2003年第14期59-62,共4页
Modern Electronics Technique
