摘要
传统图像配准技术受限于严苛的初始输入已难以满足人们的需求,近年来,视差图像的配准逐渐成为图像拼接技术中的研究热点。基于视差形成原理介绍了视差图像配准的难点与一般流程。主要研究了基于特征的视差图像配准技术,对近年来视差图像配准的研究成果进行了归纳梳理,并从基于多平面对齐的图像配准、基于网格变形的图像配准以及缝合线驱动的图像配准三个方面进行阐述。通过对各种典型的视差图像配准算法的算法思想、特点以及局限性进行描述和比较,提供该领域研究现状的系统综述,并对视差图像配准技术的研究趋势进行了展望。
Limited by the initial input images,traditional image registration technology cannot meet people’s needs.Nowadays parallax image registration has gradually become a research focus in image mosaic technology.The difficulty and general flow of parallax image registration are introduced based on the principle of parallax formation.And the featurebased parallax image registration technology is mainly introduced.The research results of parallax image registration in recent years are summarized and expanded from the image registration based on multi-plane alignment,the image registration based on mesh deformation and the seam-driven image registration.Through summarizing and comparing the algorithm ideas,characteristics and limitations of various typical parallax image registration algorithms,a systematic overview of the research status in this field is provided.Furthermore,the future research directions of parallax image registration are also prospected.
作者
夏丹
周睿
XIA Dan;ZHOU Rui(School of Information Technology in Education,Central China Normal University,Wuhan 430079,China)
出处
《计算机工程与应用》
CSCD
北大核心
2021年第2期18-27,共10页
Computer Engineering and Applications
基金
国家自然科学基金青年科学基金(61907021)。
关键词
视差图像
图像配准
图像拼接
网格变形
缝合线
parallax image
image registration
image stitching
mesh deformation
seam
