测绘学报 ›› 2018, Vol. 47 ›› Issue (1): 91-101.doi: 10.11947/j.AGCS.2018.20170129

• 摄影测量学与遥感 • 上一篇    下一篇

相同侧视方向异轨SAR图像差异化约束连接点提取

靳国旺, 熊新, 张红敏, 徐青, 刘辉, 王新田   

  1. 信息工程大学, 河南 郑州 450001
  • 收稿日期:2017-03-20 修回日期:2017-07-09 出版日期:2018-01-20 发布日期:2018-02-05
  • 作者简介:靳国旺(1977-),男,博士,副教授,博士生导师,研究方向为雷达摄影测量。E-mail:guowang_jin@163.com
  • 基金资助:
    国家自然科学基金(41071296;41474010;61401509);信息工程大学优秀青年基金(2016603502)

Tie Point Extraction for SAR Images with Same Side-looking Direction from Different Trajectories Based on Differential Constraints

JIN Guowang, XIONG Xin, ZHANG Hongmin, XU Qing, LIU Hui, WANG Xintian   

  1. Information Engineering University, Zhengzhou 450001, China
  • Received:2017-03-20 Revised:2017-07-09 Online:2018-01-20 Published:2018-02-05
  • Supported by:
    The National Natural Science Foundation of China (Nos. 41071296;41474010;61401509);The Outstanding Youth Foundation of Information Engineering University (No.2016603502)

摘要: 为了实现SAR图像区域网平差时连接点的自动稳健提取,面向相同侧视方向、近似平行轨道的异轨SAR图像,针对其方位向相对几何畸变较小、距离向相对几何畸变较大的特点,提出一种采用差异化约束的SAR图像连接点提取方法。该方法在构建SAR图像影像金字塔的基础上,采用增大方位向边长的长方形匹配窗口由金字塔顶层向下逐层进行相关系数匹配,利用方位向强约束、距离向弱约束的差异化随机采样一致性算法剔除误匹配点,并利用方位向全局双线性变换模型与距离向局部双线性变换模型进行下层金字塔影像的匹配点位预测。采用Envisat ASAR图像和国产机载SAR图像分别进行了连接点提取试验,验证了本文方法的有效性。

关键词: 合成孔径雷达, 连接点, 匹配, 提取, 差异化约束

Abstract: In order to extract tie points for SAR image block adjustment automatically and steadily,a method of tie point extraction based on differential constraints for SAR images with same side-looking direction from different trajectories is presented.It is aiming at SAR images with same side-looking direction and approximately parallel tracks,whose relative geometric distortion is small in azimuth and large in range.Firstly,image pyramids are built for two SAR images,and then these images are matched with correlation coefficient calculated by rectangular window with increased azimuth side from the top layer to the bottom layer.Mismatched points are removed by RANSAC algorithm with differential constraints.Coordinates for points in lower pyramid images are estimated with global bilinear transformation model in azimuth and local bilinear transformation model in range.Experiments performed on Envisat ASAR images and Chinese airborne SAR images validated the proposed method.

Key words: synthetic aperture radar, tie points, matching, extraction, differential constraints

中图分类号: