计算机工程与应用 ›› 2021, Vol. 57 ›› Issue (11): 193-201.DOI: 10.3778/j.issn.1002-8331.2003-0249

• 图形图像处理 • 上一篇    下一篇

Worldview 3全色与短波红外影像两步式融合框架

保盈,王雷光,郭梦晓,代沁伶,郑晨   

  1. 1.西南林业大学 林学院,昆明 650224
    2.西南林业大学 大数据与人工智能研究院,昆明 650224
    3.西南林业大学 国家林业与草原局森林生态大数据重点实验室,昆明 650224
    4.西南林业大学 艺术与设计学院,昆明 650224
    5.河南大学 数学与统计学院,河南 开封 475004
  • 出版日期:2021-06-01 发布日期:2021-05-31

Two-Step Framework for Fusing Worldview 3 Panchromatic and Shortwave Infrared Bands

BAO Ying, WANG Leiguang, GUO Mengxiao, DAI Qinling, ZHENG Chen   

  1. 1.College of Forestry, Southwest Forestry University, Kunming 650224, China
    2.Institute of Big Data and Artificial Intelligence, Southwest Forestry University, Kunming 650224, China
    3.Key Laboratory of National Forestry and Grassland Administration on Forestry and Ecological Big Data, Southwest Forestry University, Kunming 650224, China
    4.School of Art and Design, Southwest Forestry University, Kunming 650224, China
    5.School of Mathematics and Statistics, Henan University, Kaifeng, Henan 475004, China
  • Online:2021-06-01 Published:2021-05-31

摘要:

Worldview 3是目前最先进的高分辨率光学卫星之一。针对Worldview 3遥感卫星数据全色波段和短波红外波段(Short-Wave Infrared,SWIR)空间分辨率差异大、波谱范围不一致所导致的融合结果块状效应以及空间分辨率增强效果有限的问题,提出一种两步式影像融合框架。该框架降低全色波段空间分辨率,实现与SWIR波段的初步融合;将初步融合的结果与原始分辨率全色波段进行第二步融合。通过选择6种典型像素级融合方法两两组合,形成36种融合方案,验证框架的可用性。选取包含植被、建筑、水体等典型地物类型的Worldview 3数据集进行实验,并采用五个定量评价指标进行评价。实验结果表明:使用两步式融合框架进行融合,通过渐进式空间细节注入的方式,避免了直接融合产生的块状效应,实现了短波红外影像的空间分辨率增强;第一步融合采用高通滤波(HPF)融合法,第二步融合采用GS(Gram-Schmidt Transform)变换融合方法,引入的空间信息最多,获得的融合结果质量最好。提出的融合框架既能避免块状效应的产生,又能有效增强SWIR波段的空间分辨率,对于其他卫星的全色与短波红外波段的融合也具有一定借鉴意义。

关键词: 影像融合框架, Worldview 3影像, 短波红外, 全色, 高通滤波融合法, GS融合法

Abstract:

Worldview 3 is one of the most advanced high-resolution optical satellites. The large spatial resolution difference and inconsistent spectral ranges between panchromatic band and Short-Wave Infrared bands(SWIR) of Worldview 3 data sets bring great challenge to the fusion. Therefore, a two-step image fusion framework is proposed to solve the problem of blocky effects and limited spatial resolution enhancement effectiveness. The framework reduces the spatial resolution of panchromatic band to achieve the first fusion with SWIR bands. The results of the initial fusion are then fused with the original resolution panchromatic band in the second step. By choosing and combining 6 typical pixel-level fusion methods in pairs, 36 fusion schemes are formed to verify the usability of the framework. The Worldview 3 data set, consisting of typical land types such as vegetation, buildings and water, is selected for the experiment, and five quantitative evaluation indexes are used for the evaluation. The experimental results show that the spatial resolution of SWIR bands is enhanced by the progressive spatial detail injection process under the proposed framework, which avoids the block effect caused by one-step fusion methods. Under the framework, when HPF(High Pass Filter) fusion method is adopted in the first fusion step, and GS(Gram-Schmidt) fusion method is adopted in the second fusion step, the most spatial information is injected and the best fusion result quality is achieved. The proposed fusion framework can not only avoid block effect, but also effectively enhance the spatial resolution of SWIR band. The framework also can be transferred to the fusion of panchromatic and short-wave infrared bands of other satellites.

Key words: image fusion framework, Worldview 3 images, short-wave infrared, panchromatic, High Pass Filter(HPF) fusion, Gram-Schmidt(GS) fusion