摘要
作为计算机视觉领域的基础技术,图像配准为高级视觉应用提供重要支撑。随着计算设备性能和视觉理论的进步,基于特征的图像配准技术迅速成为主流。为进一步深入研究图像配准方法,在归纳特征配准算法整体框架的基础上,重点阐述SIFT、SURF、Kaze及BRISK四种典型特征的主要思想,简单分类叙述特征匹配方法。实验对比分析不同特征配准算法的性能表现,并对其优缺点予以总结,探讨图像配准技术的研究改进方向。
As a basic technology in the field of computer vision,image registration provides important support for advanced vision applications.Fea⁃ture-based image registration technology has rapidly become the mainstream with the progress of computing equipment performance and vi⁃sual theory.In order to further study the image registration method,the basic framework of feature registration algorithm is summarized as the basis,the main idea of SIFT,SURF,Kaze and BRISK features are emphasized,and feature matching method is simply classified and de⁃scribed.The experiments comparatively analyzed the performance of different feature registration algorithms,summarized the advantages and disadvantages,and explored the research and improvement directions of image registration technology.
作者
姚保琛
柏春松
YAO Bao-chen;BAI Chun-song(School of Mathematics and Statistics,Fuyang Normal University,Fuyang 236037)
出处
《现代计算机》
2020年第20期52-56,共5页
Modern Computer
基金
安徽省教育厅自然科学研究重点项目(No.KJ2018A0346)
阜阳市政府-阜阳师范学院横向合作科研重点项目(No.XDHX201727)
阜阳师范大学自然科学研究重点项目(No.2018FSKJ06ZD)。
关键词
关键点检测
尺度空间
特征匹配
图像配准
Key-Point Detection
Scale Space
Feature Matching
Image Registration
