EN中文

地球信息科学学报 >

2016 , Vol. 18 >Issue 5: 599 - 605

DOI: https://doi.org/10.3724/SP.J.1047.2016.00599

遥感大数据协同计算方法

遥感影像云及云影多特征协同检测方法

  • 沈金祥 , 1, 2, * ,
  • 季漩 2
展开
  • 1. 云南国土资源职业学院 数字国土与土地管理学院,昆明 652501
  • 2. 云南省国际河流与跨境生态安全重点实验室,昆明 650091

作者简介：沈金祥(1983-),男,博士,讲师,研究方向为遥感影像处理与信息提取。E-mail:

收稿日期: 2016-03-08

  要求修回日期: 2016-03-21

  网络出版日期: 2016-05-10

基金资助

云南省应用基础研究计划项目（2013FB082）

国家自然科学基金项目（41271367）

收起

Cloud and Cloud Shadow Multi-feature Collaborative Detection from Remote Sensing Image

  • SHEN Jinxiang , 1, 2, * ,
  • JI Xuan 2
Expand
  • 1. Department of Digital Land and Land Management, Yunnan Land and Resources Vocational College, Kunming 652501, China
  • 2. Yunnan Key Laboratory of International Rivers and Transboundary Eco-Security, Kunming 650091, China
*Corresponding author: SHEN Jinxiang, E-mail:

Received date: 2016-03-08

  Request revised date: 2016-03-21

  Online published: 2016-05-10

Copyright

《地球信息科学学报》编辑部 所有
Fold

摘要

遥感影像中云及云影不同程度地影响着地物信息的有效获取。随着多源遥感数据的日益丰富,交叉应用多源、多时相遥感影像复原云及云影区的影像,以有效地获取地类演变过程是遥感大数据应用研究的重要内容。高精度的云及云影检测是遥感影像云及云影区修复的前提和保障。复杂多变的光谱特征以及难以有效表达的空间形态特征,使云及云影一直存在检测过程复杂、适用性差和精度不高的问题,难以形成稳定有效的检测方法。在对厚云、薄云、冰雪及其他地类多光谱特性分析的基础上,本文提出了一种云及云影的多特征协同检测方法。首先,对冰雪、云及其他地物类型可分性较好的红、短波红外、热红外波段,利用SAM方法匹配云光谱特征曲线,并进一步结合短波红外波段像元绝对值区分云与冰雪,以及热红外波段像元绝对值区分云及其他地物类型;其次,通过组合云影定向移动模型与近红外波段亮度阈值检测出云影像元。对具备这些光谱波段的Landsat-8进行实验,结果表明多光谱曲线、“诊断性”波段及空间关系多特征耦合能有效地检测出影像中的薄云、厚云及云影,整体精度优于95%。

关键词： 云检测; 云影检测; 多特征协同; 阈值

本文引用格式

沈金祥 , 季漩 . 遥感影像云及云影多特征协同检测方法[J]. 地球信息科学学报, 2016 , 18(5) : 599 -605 . DOI: 10.3724/SP.J.1047.2016.00599

Abstract

Cloud and its shadow have certain degrees of impacts on the information extraction from remote sensing images. As the multi-source remote sensing data has become increasingly abundant in recent years, the cross application of the multi-source and multi-temporal remote sensing image for restoring the cloud and its shadow region, and for effectively acquiring the change information for the ground objects is an important content in studying the application of remote sensing big data. The precise detection of cloud and its shadow information is the premise and guarantee of their restoration. In general, the cloud and cloud shadow detection methods always use their spectral or spatial shape and the texture characteristics as references. However, regarding the complex and changeable spectrum and the inexpressible spatial shape characteristics, the cloud and cloud shadow information have always been difficult to be effectively detected. Based on the analysis of the spectral characteristics of thick clouds, thin clouds, snow and ice, and other feature types, a cloud and cloud shadow multi-feature collaborative detection method was proposed. (1) First of all, the cloud detection is executed. The proposed method extracts the standard thick cloud spectrum curve from the reflectance-calibrated image. Afterwards, the SAM method is used to match the cloud spectral curve from the distinguishable (red, shortwave infrared, thermal infrared) bands combination, with the absolute value of the shortwave infrared band pixel integrated to distinguish between cloud and snow, and the absolute value of the thermal infrared band pixel used to distinguish between cloud and other types of ground objects. (2) Next, the cloud shade detection is performed. Firstly, we expand the detected cloud pixel border, and produce a potential shadow mask area. Afterwards, we move the potential shadow mask along the direction of sun radiation to some distance. Thirdly, we detect the cloud shadow pixels using the brightness threshold of the near infrared band within the moved potential cloud shadow mask area. After several moves of the potential shadow mask and the implementations of cloud shadow detection based on the infrared band brightness threshold, eventually a complete cloud shadow mask is produced. The LANDSAT-8 image having the above mentioned bands is adopted in an experiment and the experimental result shows that the combination of spectral curve, "diagnosis" band and spatial relationship features can effectively detects the thin clouds, thick clouds and cloud shadows from the multispectral remote sensing image, and the overall accuracy is higher than 95%.

Key words： cloud detection; cloud shadow detection; multi-feature collaborative; threshold

1 引言

在遥感成像过程中,由于云及云影的遮挡,使影像中相应区域的地物信息难以被有效获取,这是遥感（尤其是卫星遥感）影像应用中一直存在的问题。近年来,随着不同平台的遥感数据日益丰富,利用遥感数据间的互补性,在高精度的云及云影区定位数据的支持下,通过多源、多时相遥感影像复原云及云影区的影像信息,促进遥感影像信息表达的完整性和持续性是遥感大数据分析与应用的重要研究内容。
遥感数据发布方一般通过特定算法[1-2]得到粗略的云信息,并在发布数据时提供有参考云量覆盖比例说明,甚至在质量评价影像文件中也有云信息,但其精度仍达不到实际应用需求。目前,众多学者对云及阴影的检测与修复开展了研究,可将云及云影的检测方法分为2大类：（1）基于光谱特征的检测方法,对单期影像应用可见光与红外波段的光谱特性[3-8],或对多时相影像应用光谱时间变异特性[9-15],通过光谱特征阈值（指数）检测云信息;（2）基于云的形态纹理特征的检测方法,通过对云的典型形态与纹理特征的表达、计算来检测云信息[16-18]。此外,也有部分综合光谱与空间形态、纹理特征进行云检测研究[19-20]。然而,对于遥感影像中的厚云、薄云、卷云、云影而言,空间形态与光谱特征各异,尤其薄云除有云信息外还混合有下垫面信息,因此,难以通过简单的光谱或空间形态特征实现云及云影信息的稳定、有效检测。多时相影像光谱特征的变异性分析虽然容易检测出云及云影信息,但也会引入其他真实地类发生光谱变化带来的干扰,同样需要加入判别规则进行排除。
在多维特征空间中对遥感影像像元（或像元簇、对象）进行监督分类与非监督分类时,一般特征越多则各类别间的可分性越强、分类精度越高。然而,相比于多特征使用复杂的分类与识别算法,进行单目标类别（即专题）信息的提取时,选择性地应用具有“诊断性”的少数几个特征,对于算法的稳定性和可靠性来说更为重要。对植被、裸地、不透水地表、水域、冰雪、云、阴影等典型地类光谱特征进行统计分析可发现,尽管在连续的多波段光谱曲线上反映出的薄、厚云的光谱特征各异,但仅从其中的几个波段上看,它们表现出一些区别于背景地类的共性;另外,云影虽与其他阴影或其他暗目标各波段亮度值非常相似,但它却与云存在太阳照射方向上的空间映射关系。基于这2点,本文提出了一种云及云影的多特征协同检测方法。首先,对于冰雪、云及其他地物类型可分性较好的红、短波红外、热红外波段特征,利用SAM方法匹配云光谱特征曲线,并进一步结合短波红外波段像元绝对值区分云与冰雪,以及热红外波段像元绝对值区分云及其他地物类型;其次,通过组合云影定向移动模型与近红外波段亮度阈值检测出云影像元。

2 研究方法

云及云影的多特征协同检测包括影像预处理、云检测、云影检测3个步骤。最后输出与原始影像相对应的云及云影掩膜影像（图1）。
Fig. 1 Flowchart of cloud and cloud shadow multi-feature collaborative detection

图1 云及云影多特征协同检测流程

2.1 影像预处理

影像预处理主要通过对原始影像像元量化值（DN值）进行计算,形成能准确表达地物反射率、亮度温度以及各波段光谱关系的新影像数据。影像预处理包括对可见光及近红外波段的大气顶层（TOA）反射率计算,以及热红外波段传感器接收端的温度亮度计算,包括2个步骤：（1）传感器端辐射量计算,即将影像DN值转换为传感器端接收到的辐射量值;（2）可见光、近红外及短波红外波段TOA反射率以及热红外波段亮度温度反演,即将传感器端的辐射量在成像时的日地距离、太阳高度角、大气外层平均太阳辐射量值等参数支持下进一步转换为TOA反射率,对热红外波段而言,则是将传感器端的辐射量转换为亮度温度值。2个步骤所需参数一般记录在数据头文件中,具体计算方法和步骤参考文献[21]。由于大气校正过程较为复杂且对云检测的影响不大,故无需进行进一步的大气校正处理。
此外,TOA反射率与亮度温度处于不同的数量级,为便于后续波段组合分析,将反射率波段值（取值0-1）及与热红外亮度温度波段（通常取值大于200）通过乘性或加性因子转换到相同的取值范围（如字节型值域0-255）。

2.2 云检测

在遥感影像上,云及云影由于其独特颜色与空间形态特征而较易判读。然而,云及云影的目视判读依据却难以表达为计算机容易识别的特征,只有通过一些易于表达、计算且稳定可靠的特征,才能实现计算机检测的自动化。图2（a）为冰雪、厚云、薄云及其他地类的光谱曲线对比（取值拉伸至0-255）,由此可看出：（1）在可见光波段及近红外波段,云及冰雪具有均较高的电磁波反射率;（2）在短波红外波段,云具有比冰雪更高的反射率值,这主要是冰雪对1.3~3 μm波长范围内的电磁波具有强吸收作用;（3）在反映温度的8~14 μm波长范围内的热红外波段,云及冰雪由于具有相对较低的温度值而呈现出较低的亮度值。
Fig. 2 The spectral curves of snow-ice, thick cloud, thin cloud and other feature types

图2 冰雪、厚云、薄云及其他地类光谱曲线对比

从图2（a）可看出,单一波段阈值难以实现薄、厚云与冰雪及其他地类的有效区分。另外,由于高维特征空间中相似性判别的多解性,在6个特征波段空间中即使应用复杂分类算法也同样难以区分出这4个类别。然而,通过引入几个典型的“诊断性”特征波段则较容易将薄云、厚云检测出来。在可见光-近红外波段以及热红外波段,云及冰雪具有相似且能够显著区别于其他地类的光谱特性,而在短波红外波段云与冰雪则有具有较大的可区分性。为此,结合可见光-近红外、短波红外及热红外波段特征即可实现云的检测。为降低特征维数,可以选择可见光-近红外中的一个波段与短波红外、热红外组合进行云检测。蓝波段受大气影像较为严重,近红外波段由于植被的高反射率也会与薄云部分重叠,故红、绿波段是较好的选择。
图3为冰雪、厚云、薄云及其他地类假彩色合成影像。由红、短波红外及热红外波段4种地类的光谱曲线（图2（b））及假彩色合成图像（图3（b））对比可看出,这3种特征即为云检测的“诊断性”特征。尽管这3个波段的绝对值不尽相同,但它们的光谱曲线形状非常相似,通过红、热红外波段的降与升这一特性即可将云、冰雪与其他地物类别进行区别。在遥感影像监督分类中,光谱角制图（SAM）[22]用于相似光谱曲线形状目标间的匹配,适合用于云及冰雪的检测。通过红、短波红外、热红外波段合成影像中选择相对稳定的厚云光谱作为标准光谱逐像元计算影像SAM值,再通过一定的阈值可检测出云信息。厚云在计算出的SAM影像上具有较小的值,较容易通过设置阈值检测出来。然而,薄云与冰雪及其他类别在SAM影像会有一定的重叠,为进一步提高薄云的检测精度,除通过SAM光谱曲线匹配从相对波段取值上进行识别外,再加入短波红外及热红外波段绝对阈值作为判别依据。整个云检测规则如式（1）所示。
cosα = ∑ i = 1 nb t i r i ( ∑ i = 1 nb t i 2 ) 1 2 × ( ∑ i = 1 nb r i 2 ) 1 2 < T 1 B Swir < T 2 B T h er m al < T 3 （1）
式中：nb为波段数目（取值为3）;ti,ri分别为待判别像元及参考类别（厚云）在第i波段（即红、短波红外及热红外3个波段）取值;T1为待判别像元光谱与参考类别的SAM（相似性）阈值;T2为短波红外波段阈值,用于区分厚云与冰雪;T3为热红外波段阈值,用于进一步区分薄云与其他高反射地物类别。T1、T2、T3为反复实验后选择的经验值。其中, cosα , T 1 ∈ [ 0,1 ] ; T 2 , T 3 ∈ 0,255 。
Fig.3 False color composite images

图3 冰雪、厚云、薄云及其他地类假彩色合成影像

2.3 云影检测

遥感影像中的阴影一般有山体阴影、建筑物阴影、云阴影、植被阴影等。阴影为弱信息区或无信息区,表现为较低的像元值,这与水体等低反射率地类光谱特征非常相似,难以通过简单的光谱特征将其区分。对于云影来说,其大小、形状、位置除了与云本身相关外,还受云高、太阳高度、方位、传感器位置等因素影响。通常参考云影与云的空间关系以及相似的形状,将云影于其他同样低反射率的地物类别进行区别。其中,形状这一特征的有效应用在于云及云影的高精度分割以及形状特征的有效表达,并且也依赖于空间关系来寻找候选云影区。太阳方位角α决定了云影相对云本身的方位,太阳高度角和云高则决定了云影离开云本身的距离。然而,影像中不同的云高度不尽相同,云几何中心点的成像倾角也不相同,这就难以通过几何计算解决云影的精确定位问题。为简化云影检测,可通过在阴影方向上一定距离范围内寻找云影像元来解决。由于云影与云形状相似,可对上述检测出的云影像向外膨胀扩展若干个像元作为缓冲区,然后按照太阳方位角指向以一定步长（Ds）整体移动一定距离（Dmax）,移动一步就在候选云影区寻找一次云影像元（图4）。阴影在每个波段上都呈现出较低的像元值,但在近红外（Nir）波段上与其他地物对比更为明显,故本文选择Nir波段用于云影亮度阈值判断。
Fig.4 Cloud shadow detection procedure

图4 云影检测流程图

3 实验结果与分析

3.1 实验数据

云及云影的检测涉及红、近红外、短波红外、热红外4个波段,方法独立于研究区及数据源,只要具备这4个波段的遥感影像都可成功实现云及云影的检测。本文以目前在轨运行稳定且运用广泛的Landsat 8 OLI卫星影像作为实验数据,选取位于滇西北中甸县附近云量较多、覆盖冰雪、高反射裸地、低反射水域的1景影像（轨道号为132041,文件标识号为LC81320412013115LGN01,成像时间为2013-04-25,参考云量为23.27%）用于开展检测实验。

3.2 实验结果分析

云及云影的检测在ENVI/IDL环境下实现。在预处理阶段,首先对可见光、近红外、短波红外波段计算为取值0-1间的TOA反射率后,通过乘以255将其拉伸至0-255字节型数据类型;然后,选取波长中心波长为10.9的第一个热红外波段转换为传感器端亮度温度值,经统计其有效像元亮度值为250.91-318.39,仅需减去250即可将其转为0-255字节型数据类型。此外,还需对参与云及云影检测的各波段（或波段合成彩色影像）进行一次5×5的低通滤波模糊化处理,以消除个别噪声像元及边界过渡区混合像元的干扰,并形成较为平滑的边界。
对红、短波红外、热红外3个波段合成假彩色影像,以厚云光谱（{243,166,13}）逐像元计算SAM值,再利用SAM阈值0.55、短波红外波段阈值60及热红外波段阈值40进行云检测,对检测边界向外扩展2个像元以将云过渡区包围进来。将预处理后的影像通过集成于ENVI中的Fmask算法进行云检测（核大小为5,云检测概率阈值为默认值22.5）,以及对Landsat 8附带数据质量说明文件（BQA）中云标识为“yes”或“maybe”的像元检测出来进行对比。
从图5的云检测效果对比可看出,SAM、短波红外、热红外3个特征的耦合能很好地检测出影像中的各类型云,同时,结合空间滤波与空间扩展运算,云区域边界较为平滑并且减少了破碎的云信息。FMask算法对云的边界检测相对较好,但存在冰雪区域的错误提取问题;BQA则不仅存在较多的云检测错误,云区边界也极为破碎。
Fig.5 Results of cloud detection

图5 云检测效果

通过ENVI/IDL编程,按照太阳高度角α指定的方向以及指定的云影移动步长,计算并赋予检测出的云影像左上角坐标,按空间坐标进行后续叠加即实现了云影的移动。对每次移动后的云影掩膜,根据近红外波段的阈值标识出云影像元。实验中指定最大云影距离Dmax为3 km,移动步长150 m（5个像元）,即最大移动20次后结束,近红外波段云影检测阈值为20。
由图6可看出,云影同样取得了较好的检测效果。对于较为简单的平原区,没有山体阴影、水域等同样低反射地类的干扰,主要依赖于近红外的亮度阈值的云影检测效果较好。通过云的移动限制了阴影检测的作用范围,但如果云的移动过程中碰上了水域、山体阴影,则依然会发生检测错误。多次对影像中的云及云影区域抽取随机样本进行检测的精度表明,云的检测精度优于98%,云影检测精度约为90%。
Fig.6 Results of cloud shadow detection

图6 云影检测效果

4 结论

虽然可以通过短波红外、热红外、近红外及可见光多个波段特征利用SAM或其他分类器直接检测出云及云影信息,但是由于高维空间相似性度量的不确定性,多波段的加入使SAM的可靠性随之降低。因此,本文利用多谱遥感影像中的热红外、短波红外、红波段SAM匹配特征,联合短波红外及热红外波段绝对阈值,即相对波段值与绝对波段值的联合取得较高的云检测精度。在检测过程中,SAM阈值、热红外波段阈值、短波红外阈值及近红外波段阈值都影响着云及云影的检测效果,本文通过经验阈值取得相对较优的结果。如何“因图而异”自适应地确定这些阈值是遥感影像云及云影检测实现自动化、批量化运行的关键。此外,通过太阳照射方向匹配云影尽管也取得了一定的效果,通过改进移动过程中的匹配策略预期能取得更为理想的效果。自适应阈值选择及云影匹配模型是提高精度、实现批量化的关键,也为其他地物类型提取提供了研究思路。

The authors have declared that no competing interests exist.

参考文献

原文顺序 | 文献年度倒序 | 文中引用次数倒序
[1]
Irish R.Landsat-7 automatic cloud cover assessment[C]. Proceedings of SPIE 4049 Algorithms for Multispectral, Hyperspectral, and Ultraspectral Imagery, 2000:348-355.

[2]
Irish R, Barker J L, Goward S N, et al.Characterization of the Landsat-7 ETM+ Automated Cloud-Cover Assessment (ACCA) algorithm[J]. Photogrammetric Engineering and Remote Sensing, 2006,72(10):1179-1188.

[3]
Zhu Z, Woodcock C E.Object-based cloud and cloud shadow detection in Landsat imagery[J]. Remote Sensing of Environment, 2012,118(15):83-94.A new method called Fmask (Function of mask) for cloud and cloud shadow detection in Landsat imagery is provided. Landsat Top of Atmosphere (TOA) reflectance and Brightness Temperature (BT) are used as inputs. Fmask first uses rules based on cloud physical properties to separate Potential Cloud Pixels (PCPs) and clear-sky pixels. Next, a normalized temperature probability, spectral variability probability, and brightness probability are combined to produce a probability mask for clouds over land and water separately. Then, the PCPs and the cloud probability mask are used together to derive the potential cloud layer. The darkening effect of the cloud shadows in the Near Infrared (NIR) Band is used to generate a potential shadow layer by applying the flood-fill transformation. Subsequently, 3D cloud objects are determined via segmentation of the potential cloud layer and assumption of a constant temperature lapse rate within each cloud object. The view angle of the satellite sensor and the illuminating angle are used to predict possible cloud shadow locations and select the one that has the maximum similarity with the potential cloud shadow mask. If the scene has snow, a snow mask is also produced. For a globally distributed set of reference data, the average Fmask overall cloud accuracy is as high as 96.4%. The goal is development of a cloud and cloud shadow detection algorithm suitable for routine usage with Landsat images.

DOI

[4]
Zhu Z, Wang S X, Woodcock C E.Improvement and expansion of the Fmask algorithm: cloud, cloud shadow, and snow detection for Landsats 4-7, 8, and Sentinel 2 images[J]. Remote Sensing of Environment, 2015,159:269-277.

[5]
李微,方圣辉,佃袁勇,等.基于光谱分析的MODIS云检测算法研究[J].武汉大学学报·信息科学版,2005,30(5):435-439.总结了云、雪、植被、沙漠、水体等目标的光谱特性以及云和不同目标间的光谱差异,结合MODIS影像的波段范围,提出了一种对不同下垫面通用的多光谱云检测算法,试验证明了算法的有效性。

DOI

[ Li W, Fang S H, Dian Y Y, et al.Cloud detection in MODIS data based on spectrum analysis[J]. Geomatics and Information Science of Wuhan University, 2005,30(5):435-439. ]

[6]
Kong X S, Qian Y G, Zhang A D.Cloud and shadow detection and removal for Landsat-8 data[C]. Proceedings of SPIE 8921, MIPPR 2013: Remote Sensing Image Processing, Geographic Information Systems, and Other Applications, 2013:89210N.

[7]
张宁,谢宝蓉,沈霁,等.一种基于光谱反射率特性的星载高光谱图像云检测方法[P].上海:上海航天测控通信研究所,2014.

[ Zhang N, Xie B R, Shen W, et al.A spectral reflectance characteristics based cloud detection method for satellite hyperspectral image[P]. Shanghai Aerospace Measurement and Control Communication Institute, 2014. ]

[8]
冯书谊,张宁,沈霁,等.基于反射率特性的高光谱遥感图像云检测方法研究[J].中国光学,2015,8(2):198-204.为了提高光谱。感图像数据的利用率，提出了一种基于光谱反射率特 性的星载高光谱图像云检测方法。首先分析了各类地物在不同光谱下的反射率特征，根据其特点从星载高光谱图像中选取特定5个波长的图像数据，然后依照图像地 物反射率及阈值判定法区分出高云区和中低云区，最后对二者合并，得到目标云区。试验结果表明，该方法的云检测率可达90％以上，虚警率在1％以内，可以准 确区分云与雪地区域、沙漠区域以及植被区域，实现云检测。

DOI

[ Feng S Y, Zhang N, Shen J, et al.Method of cloud detection with hyperspectral remote sensing image based on the reflective characteristics[J]. Chinese Optics, 2015,8(2):198-204. ]

[9]
Wang B, Ono A. Automated detection and removal of clouds and their shadows from Landsat TM images[J]. IEICE Transactions on Information and Systems, 1999, E82-D(2):453-460.In this paper, a scheme to remove clouds and their shadows from remotely sensed images of Landsat TM over land has been proposed. The scheme uses the image fusion technique to automatically recognize and remove contamination of clouds and their shadows, and integrate complementary information into the composite image from multitemporal images. The cloud regions can be detected on the basis of the reflectance differences with the other regions. Based on the fact that shadows smooth the brightness changes of the ground, the shadow regions can be detected successfully by means of wavelet transform. Further, an area-based detection rule is developed in this paper and the multispectral characteristics of Landsat TM images are used to alleviate the computational load. Because the wavelet transform is adopted for the image fusion, artifacts are invisible in the fused images. Finally, the performance of the proposed scheme is demonstrated experimentally.

DOI

[10]
Tseng D C, Tseng H T, Chien C L, et al.Automatic cloud removal from multi-temporal SPOT images[J]. Applied Mathematics and Computation, 2008,205:584-600.提供台灣中央大學的博碩士論文、考古題、期刊論文、研究計畫等下載

[11]
Hagolle O, Huc D M, Pascual V, et al.A multi-temporal method for cloud detection, applied to Formosat-2, Venµs, Landsat and Sentinel-2 images[J]. Remote Sensing of Environment, 2010,114(8):1747-1755.

[12]
Jin S M, Homer C, Yang L M, et al.Automated cloud and shadow detection and filling using two-date Landsat imagery in the USA[J]. International Journal of Remote Sensing, 2013,34(5):1540-1560.Not Available

DOI

[13]
Goodwin N R, Collett L J, Denham R J, et al.Cloud and cloud shadow screening across Queensland, Australia: an automated method for Landsat TM/ETM plus time series[J]. Remote Sensing of Environment, 2013,134:50-65.The 20 + year collection and moderate resolution of Landsat Thematic Mapper (TM) imagery and Enhanced Thematic Mapper (ETM+) imagery provide a crucial data source for analysing land surface change over time for a range of applications. In Queensland, Australia, a number of government policies, natural resource management programmes and research activities are reliant on large-area, multi-temporal land cover monitoring applications based on Landsat satellite imagery. However, clouds and associated cloud shadows frequently obstruct the view of the land surface. The restriction of analyses to cloud-free imagery will reduce the opportunities to sample the land surface and limit the analysis of trends in reflectance overtime. This study presents a new automated method to screen cloud and cloud shadow and is intended for application to entire time series of Landsat imagery rather than single images processed in near- real time. The method uses a hierarchical approach and takes advantage of spectral, temporal, and contextual information. Outliers are located relative to the time series of land surface reflectance by smoothing time series information using minimum and median filters which are then used in multi-temporal image differencing. Seeded region grow and morphological dilation (pixel buffering) filters are then applied to map a larger spatial extent of the cloud/cloud shadow. Spectral and contextual rules were developed empirically using calibration and validation data derived from six Landsat WRS Path/Rows (number of images = 60) with varying climatic and land surface characteristics across the state of Queensland, Australia. The validation demonstrates that cloud contaminated pixels were accurately classified with producer's, user's and overall accuracies of 98, 87 and 97%, respectively. The ability to detect cloud shadow was less accurate, in comparison, with producer's, user's and overall accuracies of 90, 62 and 97%, respectively. The pixel buffer was found to be the largest source of commission error for cloud and cloud shadow in the final classifications. However, for many applications removing additional cloud/shadow at the expense of higher commission errors may be desirable. The performance of the method was also compared with the published Fmask (Function of mask) method. This demonstrated a moderate improvement in the detection of cloud (producer's accuracies: time series 98% and Fmask 90%; and an equivalent user's accuracy of 87%), and a significant improvement in the detection of cloud shadow (producer's accuracies: time series 90% and Fmask 78%; user's accuracies: time series 62% and Fmask 50%). Importantly, the results indicate that this automated method is robust and that temporal information can improve the detection of cloud and cloud shadow, although shadow detection above cropping areas is limited. The calibration/validation of the method has been restricted to Queensland, Australia. With further development there is potential for this method or one using a similar framework to have wider application in other landscapes. Crown Copyright (c) 2013 Published by Elsevier Inc. All rights reserved.

DOI

[14]
Tang H, Yu K, Hagolle O, et al.A cloud detection method based on a time series of MODIS surface reflectance images[J]. International Journal of Digital Earth, 2013,6:157-171.This study employs visible-near infrared and short wave infrared datasets of Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) to map salt diapirs and salt diapir-affected areas using Multi-Layer Perceptron (MLP) in the Zagros Folded Belt, Iran, and introduces the role of earth observation technology and a type of digital earth processing in lithological mapping and geo-environmental impact assessment. MLP neural network model with several learning rates between 0.01 and 0.1 was carried out on ASTER L1B data, and the results were compared using confusion matrices. The most appropriate classification image for L1B input to MLP was produced by learning rate of 0.01 with Kappa coefficient of 0.90 and overall accuracy of 92.54%. The MLP result of input data set mapped lithological units of salt diapirs and demonstrated affected areas at the southern and western parts of the Konarsiah and Jahani diapirs, respectively. Field observations and X-ray diffraction analyses of field samples confirmed the dominant mineral phases identified remotely. It is concluded that MLP is an efficient approach for mapping salt diapirs and salt-affected areas.

DOI

[15]
Zhu Z, Woodcock C E.Automated cloud, cloud shadow, and snow detection based on multitemporal Landsat data: an algorithm designed specifically for land cover change[J]. Remote Sensing of Environment, 2014,152:217-234.

[16]
Li P F, Dong L M, Xiao H C, et al.A cloud image detection method based on SVM vector machine[J]. Neurocomputing, 2015,169:34-42.The satellite remote sensing image data volume is too big, therefore, transmitting, storing and processing mass data is very difficult. Thus, the current methods may not perform well. In this paper, we propose a cloud image detection method based on SVM vector machine to remove thick cloud data to reduce the amount of data to improve the efficiency of the data. Firstly, the satellite remote sensing image is divide into small blocks, and the brightness characteristics of the sub-block image is extracted to accomplish the preliminary detection. Then the average gradient and the angle of the gray level co-occurrence matrix second-order moment for sub-block image based on the texture features of the sub-block image is calculated as the basic of SVM victor machine. The sub-block cloud image is used as learning samples of the SVM classifier that has brightness characteristics, and the classification model is obtained from the training of the SVM classifier to realize a detail classification of the cloud image detect based on the SVM victor machine. Finally, we conduct experiments on cloud image detection method based on SVM vector machine. Experiment results demonstrate detection accuracy of the method proposed could reach above 90%.

DOI

[17]
王奎,张荣,尹东,等.基于边缘特征和AdaBoost分类的遥感图像云检测[J].遥感技术与应用,2013,28(2):263-268.<p>传统遥感图像云检测方法在处理山地、雪地、暗云等场景时极易发生错判,准确度较低。通过对遥感图像中云与地物的不同特点进行分析,提出一种新的遥感图像边缘特征描述方法,结合图像的边缘特征和灰度特征使用AdaBoost分类器进行云图分类,并利用图像的空间相关性对分类结果进行修正。经10万余幅图像测试结果表明:该算法与传统算法相比准确度极大提高,正确率达到96%以上,且运算速度快,满足实时性要求。</p>

[ Wang C, Zhang R, Yin D, et al.Cloud detection for remote sensing image based on edge features and AdaBoost classifier[J]. Remote Sensing Technology and Application, 2013,28(2):263-268. ]

[18]
陈振炜,张过,宁津生,等.资源三号测绘卫星自动云检测[J].测绘学报,2015,44(3):292-300.<p>光学卫星遥感影像自动云检测是卫星产品生产系统的一个重要环节.利用资源三号卫星编目生成的浏览图,采用树状判别结构进行云检测,对浏览图进行分块,提取子块图像的特征进行云地分类.由于云类和地物类过于繁杂,且浏览图的分辨率较低,传统通过图像特征对云地进行分类的算法有很大的局限性,针对这一问题,本文提出了在分类之前对原始的子块图像进行增强处理,扩大云和地物的纹理差异,然后以二阶矩、一阶差分等作为云地分类的图像特征,并在多尺度空间内进行特征延拓,经过综合分析估计云在影像中的比例.该云检测算法应用于资源三号卫星应用系统工程,实际测试结果表明,该算法能够较好地提升云量检测的准确率.</p>

DOI

[ Chen Z W, Zhang G, Ning J S, et al.An automatic cloud detection method for ZY-3 satellite[J]. Acta Geodaetica et Cartographica Sinica, 2015,44(3):292-300. ]

[19]
马雷,张秀玲,田原,等.一种基于空间与光谱信息的高光谱自动云检测方法[P].北京:中国科学院自动化研究所,2012.

[ Ma L, Zhang X L, Tian Y, et al.An automatic cloud detection method for hyperspectral image based on spatial and spectral information[P]. Institute of Automation, Chinese Academy of Science, 2012. ]

[20]
秦雁,邓孺孺,何颖清,等. 基于光谱及几何信息的TM图像厚云去除算法[J].国土资源遥感,2012(4):55-61.为了去除厚云及其阴影对光学遥 感图像的影响,以TM图像为实验数据提出一种基于光谱及几何信息的去厚云算法。在分析单图像云区多光谱特征及对比多时相图像光谱特征的基础上,首先检测光 谱特征明显的厚云,依据云和云影成对出现的成像几何关系,按确定的方位和距离搜索云影;然后采用数学形态学的侵蚀与膨胀算法对云区边缘进行碎片去除及填 补、合并处理,使其准确反映图像中受云影响的数据总量;最后利用光谱线性回归匹配后的参考图像替换目标图像中的云区,达到去云目的。实验结果表明,上述算 法去厚云效果显著,能够有效排除水体及地形阴影对云影识别的影响,具有快速、简单、实用性强的特点。

DOI

[ Qin Y, Deng R R, He Y Q, et al.Algorithm for removing thick clouds in TM image based on spectral and geometric information[J]. Remote Sensing for Land & Resources, 2012,4:55-61. ]

[21]
Chander G, Markham B L, Helder D L.Summary of current radiometric calibration coefficients for Landsat MSS, TM, ETM+, and EO-1 ALI sensors[J]. Remote Sensing of Environment, 2009,113:893-903.

[22]
Kruse F A, Lefkoff A B, Boardman J B, et al.The spectral image processing system (SIPS) - interactive visualization and analysis of imaging spectrometer data[J]. Remote Sensing of Environment, 1993,44:145-163.The Center for the Study of Earth from Space (CSES) at the University of Colorado, Boulder, has developed a prototype interactive software system called the Spectral Image Processing System (SIPS) using IDL (the Interactive Data Language) on UNIX-based workstations. SIPS is designed to take advantage of the combination of high spectral resolution and spatial data presentation unique to imaging spectrometers. It streamlines analysis of these data by allowing scientists to rapidly interact with entire datasets. SIPS provides visualization tools for rapid exploratory analysis and numerical tools for quantitative modeling. The user interface is X-Windows-based, user friendly, and provides 鈥減oint and click鈥 operation. SIPS is being used for multidisciplinary research concentrating on use of physically based analysis methods to enhance scientific results from imaging spectrometer data. The objective of this continuing effort is to develop operational techniques for quantitative analysis of imaging spectrometer data and to make them available to the scientific community prior to the launch of imaging spectrometer satellite systems such as the Earth Observing System (EOS) High Resolution Imaging Spectrometer (HIRIS).

DOI

Options
文章导航

/

〈 〉