项目名称: 面向多场景图像自适应的三维人脸建模研究
项目编号: No.61202191
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 计算机科学学科
项目作者: 龚勋
作者单位: 西南交通大学
项目金额: 23万元
中文摘要: 基于图像的三维人脸建模是众多相关应用的核心支撑技术。本项目以无需用户配合的三维人脸建模为背景,以多场景、异源人脸图像为研究对象,以机器学习模型、人脸局部特征合成分析、光照模型为主要手段,挖掘多视图、异构图像的内在本质联系,建立人脸图像表观与其几何形状的相关关系,研究可扩展、自适应的三维人脸建模。本项目将提出基于不同场景人脸图像的三维人脸建模的特色理论及具体方案。研究内容及关键问题包括:(1)多场景人脸图像的典型人脸数据集构建及自适应建模策略;(2)基于多场景人脸图像的一致性建模方案;(3)可扩展的人脸建模方案,能够有效地融入新视图增强建模效果。这些研究成果将提供一套完整的基于图像的三维人脸建模理论和方法,为人脸识别等研究奠定基础。
中文关键词: 三维人脸重建;多场景;自适应;;
英文摘要: Image-based 3D face modeling is the core support technology for many applications. This project researches on modeling 3D face of noncooperative subjects from multi-scene, heterogeneous face images. Taking tools like machine learn model, face local characteristics synthesis by analysis and phone models, we aim to explore the canonical correlation between face images appearance and their geometry and develop an extensible and adapitve 3D face modeling method.We will put forward novel strategies, theory and the specific programs for 3D face modeling based on face images of multi-scene. The studies and key issues include:(1) Extract characteristic face set (CFS) from multi-scene images and propose an adaptive modeling stategy;(2) Propose an coincident modeling method for multi-scene images;(3) An extensible face modeling method, which can enhance the modeling effect through additional views. The results of these studies will provide a complete solution of image-based 3D face modeling and lay the foundation for face recognition and other research.
英文关键词: 3D face reconstruction;multiple view;adaptive;;