计算机应用 ›› 2016, Vol. 36 ›› Issue (1): 248-253.DOI: 10.11772/j.issn.1001-9081.2016.01.0248

• 虚拟现实与数字媒体 • 上一篇    下一篇

基于同态系统滤波的高分辨率遥感图像河流信息提取

杨宜菩1, 杨帆1, 潘国峰1,2, 张慧敏1   

  1. 1. 河北工业大学 电子信息工程学院, 天津 300401;
    2. 河北工业大学 微电子研究所, 天津 300401
  • 收稿日期:2015-05-28 修回日期:2015-08-08 出版日期:2016-01-10 发布日期:2016-01-09
  • 通讯作者: 杨帆(1966-),男,吉林吉林人,教授,博士主要研究方向:智能信息处理、机器视觉、传感器
  • 作者简介:杨宜菩(1991-),男,河北邢台人,硕士研究生,主要研究方向:遥感图像处理识别、智能信息处理;潘国峰(1968-),男,内蒙古呼和浩特人,副教授,博士,主要研究方向:电子信息与功能薄膜、半导体敏感器件、智能传感器系统;张慧敏(1991-),女,河北邯郸人,硕士研究生,主要研究方向:数字图像匹配与拼接、智能信息处理。
  • 基金资助:
    国家科技重大专项(2009ZX02308-004);河北省高等学校科学研究项目(Z2014088)。

River information extraction from high resolution remote sensing image based on homomorphic system filtering

YANG Yipu1, YANG Fan1, PAN Guofeng1,2, ZHANG Huimin1   

  1. 1. School of Electronic and Information Engineering, Hebei University of Technology, Tianjin 300401, China;
    2. Institute of Microelectronics, Hebei University of Technology, Tianjin 300401, China
  • Received:2015-05-28 Revised:2015-08-08 Online:2016-01-10 Published:2016-01-09
  • Supported by:
    This work is partially supported by the National Science and Technology Major Project (2009ZX02308-004) and the Science Research Programs of Hebei Educational Committee for University (Z2014088).

摘要: 为了解决在高分辨率遥感图像中易受同物异谱现象影响而导致河流信息提取不完整问题,依据河流区域在遥感图像中具有特定频率纹理信息这一特点,提出一种基于同态系统滤波的高分辨率遥感图像河流信息提取方法。该算法通过对遥感图像的频谱分析,确定河流信息在频域中的特征标识,采用一维剖面线加窗短时分析河流定位技术实现了全自动、快速全幅遥感图像河流定位和宽度估计,在同态系统下设计了低通滤波器实现对低频河流信息的提取。该算法采用"高分一号"遥感影像作为实验数据来源,相对于传统利用光谱信息的提取方法具有较高的提取精度,Kappa系数达0.85以上,并且实现全自动提取。实验结果表明所提算法能够快速、有效地提取复杂地貌区域内的河流信息。

关键词: 高分辨率遥感图像, 河流信息提取, 加窗短时分析, 同态系统滤波, 高分一号卫星

Abstract: Concerning the incomplete extraction of river information easily affected by synonyms spectrum phenomenon from the high resolution remote sensing image, considering the specific frequency texture information of the river region, a river information extraction method from the high resolution remote sensing image based on homomorphic system filtering was proposed. The spectral analysis of remote sensing image was used by the proposed algorithm to define the feature identification of river information in the frequency domain. The one-dimensional profile line adding-window short-time analysis river location technology was adopted to realize river location and width estimation of the remote sensing image automatically and quickly. Finally, the low-pass filter under the homomorphic system was designed to extract low frequency river information. The GF-1 image was used as experimental data. The extraction accuracy of the proposed algorithm was higher than the traditional method based on spectral information. The Kappa coefficient reached 0.85, and the extraction process was automatic. The experimental results show that the proposed algorithm can extract river information from complex geomorphology region quickly and effectively.

Key words: high resolution remote sensing image, river information extraction, adding-window short-time analysis, homomorphic system filtering, GF-1 satellite

中图分类号: