Abstract: This paper presents an automatic adaptation method of constructing face-model on the Candide-3 model.In the proposed methods，by different regions gray-level projecting on a face gray-level image and its edge-detection image，the parts of face are positioned approximately.In the meantime，the face’s feature points are located by using direct-locating and calculational-
locating method based on analyzing the corresponding two-value image，and the face’s three-dimension model is constructed finally.Experimental results show that this method can make the standard Candide-3 model adapting the different
faces effectively.

摘要： 以Candide-3 模型为基础，提出了一种适应不同人脸的三维建模方法。该方法通过对人脸的灰度图像及相应的边缘检测图像进行分层次区域灰度投影，确定了人脸面部器官的大致区域，同时对二值化图像进行分析，采用直接定位和计算定位两种方式，提取了人脸特征点，并以此为基础建立了人脸的三维模型。实验证明，该方法能够较好地调整标准Candide-3 模型，使之适应不同的人脸。