基于Graph Cut的图像前景提取

发布时间：2018-06-19 10:14

本文选题：交互式前景提取 + Graph　；参考：《华南理工大学》2009年硕士论文

【摘要】： 前景提取与图像合成是图像处理中的基本操作,也是视觉特效制作中是最重要和最常用的操作。在杂志、二维图形艺术、电视、广告等领域也有广泛的应用。前景提取是指从一幅图像中提取出任意形状的前景对象,本文简要介绍了图像前景提取技术的研究背景及研究现状,分析比较了各种常用算法的优缺点;重点介绍了Graph Cut算法的实现,在此基础上开发了一个简单的图像前景提取交互系统。论文的主要工作包括: 分析目前常用的各种前景提取算法,并对它们的分割效果进行了比较;详细分析了Graph Cuts算法的实现过程;研究并实现基于Graph Cut的前景提取交互系统。论文介绍了该系统的界面组成和基本使用流程、核心数据结构和算法,并对前景提取结果进行了分析评价,指出系统存在的问题和改进的方向。
[Abstract]:Foreground extraction and image synthesis are the basic operations in image processing and the most important and commonly used operations in visual special effects making. In magazines, two-dimensional graphics art, television, advertising and other fields also have a wide range of applications. Foreground extraction refers to extracting any shape foreground object from an image. This paper briefly introduces the research background and research status of image foreground extraction technology, and analyzes and compares the advantages and disadvantages of various commonly used algorithms. The implementation of Graph cut algorithm is introduced, and a simple interactive system of image foreground extraction is developed. The main work of this paper is as follows: firstly, we analyze the commonly used foreground extraction algorithms, and compare their segmentation effects, and analyze the implementation process of Graph sets algorithm in detail. An interactive system of foreground extraction based on Graph cut is studied and implemented. This paper introduces the interface composition, basic usage flow, core data structure and algorithm of the system, analyzes and evaluates the result of foreground extraction, and points out the existing problems and improvement direction of the system.
【学位授予单位】：华南理工大学
【学位级别】：硕士
【学位授予年份】：2009
【分类号】：TP391.41

【参考文献】