本文为澳大利亚莫纳什大学（作者：Paul Francis Harrison）的博士论文，共141页。
本文涉及三种图像纹理操作：从样本合成纹理，从一幅图像到另一幅图像的纹理转移，以及不完整或噪声图像的合理复原。由于人类视觉感知对纹理细节的敏感特性，因此对这些操作产生令人满意的结果是困难的。本文提出了几种实现这些操作的新方法。在纹理合成方面，本文对最佳拟合方法进行了改进，消除了该方法有时产生的“歪斜”和“垃圾”效应，同时能够快速、灵活、简单地实现，因此是一种非常实用的方法。本文提出了一种基于随机拼贴的简单快速技术，这两种技术都可以适用于将纹理从一幅图像转移到另一幅图像。接下来，提出了一种非线性预测器形式的基于图像纹理模型的噪声去除方法，该方法被应用于通过诸如调色处理(例如GIF)之类的压缩技术实现退化图像的纹理复原和高斯去噪，其性能与基于最新小波技术的方法相当。最后，基于指定形状的瓦片排列，研究了一种更抽象的纹理合成形式，用于显示最佳拟合合成的古器物原貌。
Three image texture operations areidentified: synthesis of texture from a sample, transfer of texture from oneimage to another, and plausible restoration of incomplete or noisy images. Ashuman visual perception is sensitive to details of texture, producingconvincing results for these operations can be hard. This dissertation presentsseveral new methods for performing these operations. With regard to texturesynthesis, this dissertation presents a variation on the best-fit method [Efrosand Leung, 1999, Garber, 1981] that eliminates the “skew” and “garbage” effectsthis method sometimes produces. It is also fast, flexible, and simple toimplement, making it a highly practical method. Also presented is a simple andfast technique based on random collage. Both of these techniques can be adaptedto transfer texture from one image to another. Next, a noise removal methodthat is guided by a model of an image’s texture, in the form of a non-linearpredictor, is presented. The method is applied to plausibly restoring thetexture of images degraded by compression techniques such as palettization(e.g. GIF), and to the removal of Gaussian noise, with results comparable tostate-of-the-art wavelet-based methods. Finally, a more abstract form oftexture synthesis is examined, based on the arrangement of tiles of specified shape.This is used to show the origins of the artifacts seen in best-fit synthesis.
1 引言
2 历史文献回顾
3 改进的最佳拟合纹理合成
4 引入最佳拟合纹理合成的GIMP
5 拼贴式纹理合成
6 噪声图像的合理复原
7 声明性纹理合成
8 结论与未来工作展望
附录A 测试数据及程序
附录B WSCG-2001发表的论文
下载英文原文地址：
http://page5.dfpan.com/fs/2lac8j6272d18229169/
更多精彩文章请关注微信号：