XIANG Sheng-wen, WEN Gong-jian, GAO Feng, et al. High resolution remote sensing image registration method based on multi-model representation[J]. 2015, 34(10): 22-24+28. DOI: 10.13873/J.1000-9787(2015)10-0022-03.
基于多模型表示的高分辨率遥感图像配准方法
摘要
传统的配准方法假定两幅图像之间的几何变形可以用一个统一的变换模型来描述
高分辨率遥感图像配准
尤其当图像的分辨率达到米级和亚米级时
地物高程因素产生的像点位移不容忽略
导致这些区域的变形与平坦区域不一致
难以找到一个统一的变换模型来描述整幅图像的变形。针对高分辨率图像配准中存在的实际问题
提出了一种基于多模型表示的配准方法。在初配准阶段
完成图像中大部分平坦区域的校正
建立整体模型;在精配准阶段
完成局部高程区域的校正
建立局部模型。实验结果表明:该方法是准确有效的。
Abstract
Traditional registration methods generally assume distortion between images can be described by an uniform transformation model. It’s a tough work to find a proper registration model for high resolution remote sensing images
when the resolution is up to meter or sub-meter
the pixel point displacement caused by hypsography is improper to ignore. In order to resolve the problem
a registration method for high resolution images based on multi-model representation is proposed. In the first period
adjust most of flat area and a global model is constructed; during the second period
local areas with hypsography are corrected and local models are built.Experimental result shows the proposed method is accurate and effective.
The Remote Sensing Image Matching Algorithm Based on the Normalized Cross-Correlation and SIFT [J] . Xingxing Shen,Wenxing Bao. Journal of the Indian Society of Remote Sensing . 2014 (2)