【作者】 胡亮；

【导师】 陈宇拓；

【作者基本信息】 中南林业科技大学 ， 计算机应用技术， 2010， 硕士

【摘要】 人脸图像变形动画技术是计算机动画研究的一个重要分支。无论是电影电视、游戏娱乐、远程通信还是虚拟环境,人脸动画技术都被应用其中,为人们的交流、文化和娱乐带来极大的便利和乐趣,因此,这一研究领域越来越受到人们的关注。它涉及到人体生理学、心理学、艺术、计算机图形学和视觉等等多个学科。在此领域,无数的研究者奉献了自己的心力,为我们带来更好更有表现力的人脸动画技术。本文研究的是二维人脸图像的变形动画技术。文章概括了几种常见的人脸特征提取方法,系统的介绍了各种图像变形技术和人脸动画技术,总结了图像变形的基本原理、现状研究和存在的问题。在人脸图像特征区域提取方面,通过对不同提取方法适用范围的对比,针对人脸图像结构特征和局部变形动画实现的需要,采用了积分投影与数据分析的特征区域的提取方法。在源图像到目标图像间的变形方面,提出了改进的基于特征线段的图像变形方法。这种方法先要从源图像和目标图像提取出的特征中,找出对应的特征线对,建立起两幅图像之间的映射关系。再对源图像和目标图像按一定比例进行插值,由此生成一组渐变的人脸变形图像。针对改进前的方法存在着计算量较大、运行等待时间较长的问题,本文改进后的算法运行速度有了明显提高。人脸局部变形方面,本文提出了一种基于像素点位移的变形方法,即通过一定的映射关系使图像中的像素点的像素值的位移发生改变,从而生成新的变形图像。本文以人脸局部变形中变化最明显的嘴巴和眼睛为例,研究和分析了它们的变形算法思想,提出了具体的算法。该算法生成的人脸局部变形动画连续性好、可参数化交互实现多种变形效果。本文主要就人脸图像变形动画技术的两大核心环节——人脸特征提取、图像变形展开说明,从实践的角度,论述了相关技术的实现过程。研究的方法和实验数据具备一定的实用参考价值。更多还原

【Abstract】 Human face image morphing animation technology is an important branch of the computer animation research. Facial animation technology has been applied in film and television, video games, remote communication, and virtual environment, which brings great convenience and fun in the exchange of people, cultural and entertainment. Therefore, this field of research attracts more and more attention. It involves human physiology, psychology, art, computer graphics,vision and so on. In this field, many researchers devote their effort to bring us better and more expressive human face animation technology.This article discusses morphing animation technology of 2D human face image.This article generalizes some common human face feature extraction method, introduces systematically a variety of image morphing technology and facial expression animation technology, summarizes the basic principle of image morphing, research actuality and existent problem.In the facial image feature extraction, through the comparison of application range of different extraction methods, in allusion to the configuration character of human face images and need of local animation warping actualized used in the article experiment,this article use the integral projection of extraction methods and get the feature area. In the respect of morphing between original images and target images, the article use image morphing method based on feature line after arithmetic improved. Firstly find the corresponding feature lines and establish a mapping between two images from feature points from original images and target images. Then interpolate original images and target images by a certain percentage, and this will generate a gradual morphing facial image. However, as this method before arithmetic improved costs a large amount of calculation, longer waiting time of running, the speed of method after arithmetic improved is quickly. In the respect of part facial morphing, the paper presents a morphing method based on pixel displacement, which is, through some mapping the image, pixels displacement change and generate a new image. This article takes mouth and eyes which changes the most obviously on the face as examples, and research and analyze the thought of morphing algorithm. This algorithm creat good continuity, and can preferences produce manifold morphing effort.This paper explains facial feature extraction and image morphingthe facial image morphing which are two core areas of the human face image morphing animation technology, and discusses the implementation process of related technologies.The studied method and experiment date are provided with a practical value.更多还原