（毕业设计论文）《三维人脸识别系统》.docVIP

PAGE 三维人脸识别系统研究 摘　　要 生物识别是最能体现计算机智能的特征之一，它在科研工作和日常生活应用方面都引起了巨大的轰动。在所有生物特征里面，人脸特征是最普遍和最容易获取的，所以，人脸识别一直都是模式识别和图像处理领域最热门的课题之一。为了减弱光照、表情、姿态、摄像视角、外貌(例如头发、眼睛)等变化对识别的影响，提高三维人脸识别准确性，本文针对光照、姿态、表情等影响三维人脸识别的因素进行了研究。使用基于球面谐波基图像的光照补偿算法减弱光照对人脸识别的影响，利用三维形变模型和部件技术相结合的人脸识别方法进行多姿态人脸的识别，针对表情变化带来的塑性变形采用基于泊松方程完成非中性人脸网格模型向中性模型的变形，提高带表情模型与同类中性模型之间的相似度。这三种算法在提高三维人脸识别的性能上都有一定的帮助，对于三维人脸识别系统的研究有了一定的理论基础。 关键词：人脸识别；三维人脸识别；光照；姿态；表情Abstract 关键词：人脸识别；三维人脸识别；光照；姿态；表情 Biometrics is one of the best embodies the characteristics of computer intelligence, its research work and daily life applications have caused a great sensation. In all the biological characteristics of which, facial feature is the most common and most easily accessible, so, face recognition, pattern recognition and image processing has been the hottest topic in the field. In order to weaken illumination, facial expressions, gestures, camera angle, appearance (eg hair, eyes) and so the impact of change on the recognition and improve the accuracy of three-dimensional face recognition, this paper light, gesture, facial expression and other factors affecting the 3D face recognition study. Spherical harmonics basic image-based illumination compensation algorithm reduced the influence of light on face recognition using 3D deformable model and parts is combined with multi-pose face recognition face recognition, for the expression of changes in the plastic deformation Poisson equation based on the completion of non-neutral facial mesh model deformation to the neutral model, increased with the same neutral expression model and the similarity between models. The three dimensional face recognition algorithm to improve the performance of all have some help, for 3D face recognition system for a certain theoretical basis. Key Words： Key Words：face recognition; 3D face recognition; illumination; attitude ; expression - PAGE 1 - 目　　录 TOC \o 1-3 \h \z HYPERLINK \l _Toc264672406 摘　　要 PAGEREF _Toc264672406 \h i HYPERLINK \l _Toc264672407 Abs