Research of image detection and matching algorithms

Abstract

Image matching, a fundamental computer vision method, serves as a crucial pillar for more complex vision applications. The general adoption of feature-based image registration technologies has been accelerated by advances in computing hardware and vision theory. As the current research in this field is not very sufficient, this paper gives an overview of the relevant aspects. At the beginning, this article first introduces the research background, the research achievements and the application in different fields of image feature detection and matching. The main body discusses the most current advancements in this subject, including feature points, local features, global features, matching, and optimization, after examining the classical detection algorithms from recent decades and referencing the most recent machine learning algorithm headed by depth learning, and shows the advantages and disadvantages of the algorithms. Finally, the paper summarizes and prospects the full text.