测绘学报 ›› 2016, Vol. 45 ›› Issue (6): 706-712.doi: 10.11947/j.AGCS.2016.20150375

• 摄影测量学与遥感 • 上一篇    下一篇

大孔径静态干涉成像光谱仪图像的条纹消除与配准

王府北1, 景娟娟2, 吴琼水1, 周锦松2   

  1. 1. 武汉大学电子信息学院, 湖北 武汉 430072;
    2. 中国科学院光电研究院计算光学室, 北京 100094
  • 收稿日期:2015-07-06 修回日期:2016-04-18 出版日期:2016-06-20 发布日期:2016-06-29
  • 作者简介:王府北(1992-),男,硕士,研究方向为LASIS图像处理。E-mail:277435796@qq.com
  • 基金资助:
    国家自然科学基金(61405203;61405204);国家杰出青年科学基金(61225024)

Stripes Elimenation and Registration of Large Aperture Static Imaging Spectrometer Image

WANG Fubei1, JING Juanjuan2, WU Qiongshui1, ZHOU Jinsong2   

  1. 1. School of Electronic Information of Wuhan University, Wuhan 430072, China;
    2. Key Laboratory of Computational Optics Imaging Technology, Academy of Opto-electronics, Chinese Academy of Sciences, Beijing 100094, China
  • Received:2015-07-06 Revised:2016-04-18 Online:2016-06-20 Published:2016-06-29
  • Supported by:
    National Natural Science Foundation of China(Nos.61405203;61405204);National Science Fund for Distinguished Young Scholars(No.61225024)

摘要: 大孔径静态干涉成像光谱仪获取的图像是叠加了干涉信息的二维图像,点干涉图的获取需要经过全视场的推扫,平台姿态误差会导致提取的干涉图存在误差,进而影响复原光谱准确性,因此,图像配准是干涉图提取的关键。由于图像受到干涉调制作用,导致传统的图像配准方法在应用于LASIS图像配准时配准精度有限。为了减弱干涉条纹对图像配准的影响,本文提出一种利用条纹模板消除LASIS图像条纹,再结合图像频域配准的方法实现LASIS图像的高精度配准。配准结果表明,本方法很好地消除了干涉条纹对配准精度的影响,配准精度可以达到0.0294个像素,相对于已开展的LASIS图像配准方法,沿轨方向配准精度提高了45.48%,跨轨方向配准精度提高了52.22%,图像旋转精度提高了39.13%。

关键词: 遥感, 大孔径静态干涉成像光谱仪, 配准, 亚像素, 预处理

Abstract: The images obtained by large aperture static imaging spectrometer(LASIS) are two dimensional images which contain interference information. The acquisition of the interferogram needs to pushbroom the whole field of view. The pose error of the platform can result in the error of the interferogram extraction, and then affect the accuracy of the recovered spectrum. Therefore, the image registration is the key link of the interferogram extraction. The accuracy of the existed registration methods applied to LASIS images is limited as the images are modulated by the interference stripes. In order to reduce the interference effect of the stripes in image registration, a stripe pattern is optimized in this paper to eliminate the stripes in LASIS images. Then, the frequency domain image registration method is used to realize the precise registration of LASIS image. Results show that this method eliminates the influence of interference stripes, the registration accuracy can reach up to 0.0294 pixels. Compared with the existed registration methods for LASIS, the accuracy improved 45.48% along the track direction, 52.22% in cross track direction, and image rotation accuracy improved 39.13%.

Key words: remote sensing, LASIS, registration, sub-pixel, pretreatment

中图分类号: