基于支持向量机的三维鼻子建模方法研究

摘要

This paper proposes a method of 3D nose modeling. It needs two orthogonal face photos. First, high-boost filter is needed to make the nose edge profile more clearly and so that it can then gain the nose’s 3D coordinate information better. Then, the nose is fitted by the Support Vector Machine(SVM)to model. Finally, texture mapping is applied to the model obtained by fitting in order to get a realistic 3D nose model. This method has been carried out on the platform of MATLAB and a realistic 3D nose model is successfully obtained.%提出了一种基于支持向量机的三维鼻子建模新方法，该方法采用两张正交的人脸照片，首先运用高提升滤波对整个面部照片进行处理，使鼻子边缘轮廓更加清晰，然后在鼻子区域中提取鼻子的三维坐标信息，接着用支持向量机（SVM）对鼻子进行拟合建模，最后对拟合得到的鼻子模型进行纹理贴图，得到具有真实感的三维鼻子模型。该方法已在MATLAB平台上进行了模拟，成功获得了较逼真的三维鼻子模型，并可以推广应用到三维人脸建模。