基于小波变换和子图的图像配准方法  

Study on image registration technique based on wavelet transform and subgraph

作　　者：孙鑫[1] 谢元旦[1] 任地成[1] 

机构地区：[1]辽宁科技大学计算机科学与工程学院,辽宁鞍山114051

出　　处：《计算机工程与设计》2010年第21期4653-4656,共4页Computer Engineering and Design

摘　　要：为了能够快速精确的对图像进行配准,提出了一种基于小波变换的图像配准方法。该方法对基准图像和待配准图像进行小波分解,并利用小波多分辨率分析的优势,采用由粗到精的搜索策略,减少搜索空间,提高配准精度。同时引入子图的概念,用子图代替原图作配准来降低搜索的数据量,通过高效搜索的策略和并行的计算思想来提高运算速度。实验结果表明了该方法有较好的效果,具有较高的应用价值。In order to fast and accurate image registration, an image registration method based on the wavelet transformation is proposed. In this method, the two images, i.e., the reference image and the registration image are decomposed with wavelet multi-resolution analysis method, which takes the advantages of wavelet decomposition of adopting from coarse to fine search strategy, reducing the search space and improving registration precision, with introducing subgraph concept, replaces the original image with the subgraph to make the registration, searching data quantity is reduced, high effective searching strategy and parallel computation are used to enhance the computation speed. Finally the experiment show that this method has the good effect and high application value.

关 键 词：图像配准 小波变换 子图 多分辨率分析 并行计算 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]