基于正面照的自适应三维头部网格模型研究

Self-Adaption 3D Head Gridding Model Study Based on Frontal Facial Photo

研究三维头部建模问题。针对以软件导出的三维头部模型作为标准模型受到数据量多、结构复杂的影响,对不同人脸进行匹配时需要进行数据简化操作,造成工作量大、可控性差且应用受限等问题,提出一个简单的能适应于不同人脸的三维头部模型。根据Candide-3模型提供的人脸三维信息,通过增加耳朵特征和后脑曲面模型,使用127个控制点建立了标准三维头部模型,然后配合人脸特征点信息和局部调整原理建立了与输人人脸图像相仿的三维头部模型。实验结果表明,建立的三维头部模型顶点少且具有较强真实感,可根据输入人脸图像的不同相应地产生与图像人脸相仿的形变模型,具有较好的适应性和广泛的应用价值。

Research 3D head modeling.For the 3D head model exported from some software contains a large quantity of data and has a complex structure,it must reduct the datas when use this model to match different faces.It caused heavy workload,poor controllability and the application is also limited.In this paper,a simple and self-adaption 3D head model was proposed.According to the 3D information provided by the Candide-3 model,a standard 3D head model contains 127 control data can be formed by increasing ear characteristics and hindbrain surface model. Then the information of facial feature points and local adjustment principle was used to adjust the standard 3D model to a special 3D model.The special 3D model is similar to the input facial photo.The results show that the created model has less vertexes and a strong sense of reality.It can produce different modeld according to inputting different facial photos,and it has certain adaptability and wide applicability as well.