The effect of atmospheric and topographic correction methods on land cover classification accuracy

Mapping of vegetation in mountain areas based on remote sensing is obstructed by atmospheric and topographic distortions. A variety of atmospheric and topographic correction methods has been proposed to minimize atmospheric and topographic effects and should in principle lead to a better land cover classification. Only a limited number of atmospheric and topographic combinations has been tested and the effect on class accuracy and on different illumination conditions is not yet researched extensively. The purpose of this study was to evaluate the effect of coupled correction methods on land cover classification accuracy. Therefore, all combinations of three atmospheric (no atmospheric correction, dark object subtraction and correction based on transmittance functions) and five topographic corrections (no topographic correction, band ratioing, cosine correction, pixel-based Minnaert and pixel-based C-correction) were applied on two acquisitions (2009 and 2010) of a Landsat image in the Romanian Carpathian mountains. The accuracies of the fifteen resulting land cover maps were evaluated statistically based on two validation sets: a random validation set and a validation subset containing pixels present in the difference area between the uncorrected classification and one of the fourteen corrected classifications. New insights into the differences in classification accuracy were obtained. First, results showed that all corrected images resulted in higher overall classification accuracies than the uncorrected images. The highest accuracy for the full validation set was achieved after combination of an atmospheric correction based on transmittance functions and a pixel-based Minnaert topographic correction. Secondly, class accuracies of especially the coniferous and mixed forest classes were enhanced after correction. There was only a minor improvement for the other land cover classes (broadleaved forest, bare soil, grass and water). This was explained by the position of different land cover types in the landscape. Finally, coupled correction methods showed most efficient on weakly illuminated slopes. After correction, accuracies in the low illumination zone (cos β ≤ 0.65) were improved more than in the moderate and high illumination zones. Considering all results, best overall classification results were achieved after combination of the transmittance function correction with pixel-based Minnaert or pixel-based C-topographic correction. Furthermore, results of this bi-temporal study indicated that the topographic component had a higher influence on classification accuracy than the atmospheric component and that it is worthwhile to invest in both atmospheric and topographic corrections in a multi-temporal study.     

A Cloud Removal Algorithm to Generate Cloud and Cloud Shadow Free Images Using Information Cloning

One of the main problems of optical remote sensing is clouds and cloud shadows caused by specific atmospheric conditions during data acquisition. These features limit the usage of acquired images and increase the difficulty in data analysis, such as normalized difference vegetation index values, misclassification, and atmospheric correction. Accurate detection and reliable cloning of cloud and cloud shadow features in satellite images are very useful processes for optical remote sensing applications. In this study, an automated cloud removal algorithm to generate cloud and cloud shadow free images from multitemporal Landsat-8 images is introduced. Cloud and cloud shadow areas are classified by using process-based rule set developed by using spectral and spatial features after applying simple linear iterative clustering superpixel segmentation algorithm to the image to find cloud pixel groups easily and correctly. Segmentation-based cloud detection method gives better results than pixel-based for detection of cloud and cloud shadow patches. After detection of clouds and cloud shadows, cloud-free images are created by cloning cloudless regions from multitemporal dataset. Spectral and structural consistency are preserved by considering spectral features and seasonal effects while cloning process. Statistical similarity tests are applied to find best cloud-free image to use for cloning process. Cloning results are tested with the structural similarity index metric to evaluate the performance of cloning algorithm.     

面向对象方法的时间序列MODIS数据湿地信息提取——以洞庭湖流域为例

以洞庭湖流域为研究区,对大范围湿地信息遥感提取方法进行了研究。先基于时间序列MODIS EVI及物候特征参数,通过J-M(Jeffries-Matusita distance)距离分析,构建了MODIS(250 m)最佳时序组合分类数据;其次,通过Johnson指数确定了最佳分割尺度,采用面向对象的遥感分类方法(Random tree分类器)提取了洞庭湖流域的湿地信息,并验证该方法的适用性。研究结果表明,基于时序数据与面向对象的Random tree分类的总体精度和Kappa系数分别为78.84%和0.71,较之基于像元的相同算法的总体分类精度和Kappa系数分别提高了5.79%和0.04。同时,基于面向对象方法的湿地整体的用户精度与生产者精度较基于像元方法分别提高了4.56%和6.21%,可有效提高大区域湿地信息提取的精度。     

结合像元级和目标级的高分辨率遥感影像建筑物变化检测

快速、精准的建筑物变化检测对城市规划建设等业务管理具有重要意义。随着卫星遥感技术的快速发展,基于高分辨率遥感影像的建筑物变化检测得到了广泛关注。针对像元级建筑物变化检测方法往往精度不足而目标级建筑物变化检测方法过程烦琐等问题,本文提出结合像元级和目标级的高分辨率遥感影像建筑物变化检测方法。首先综合高分辨率遥感影像的多维特征,利用随机森林分类器进行影像集分类,以获取像元级建筑物变化检测结果;然后对后时相遥感影像进行图像分割,获得影像对象;最后融合像元级建筑物变化检测结果和影像对象,识别变化的建筑物目标。利用双时相QuickBird高分辨率遥感影像进行建筑物变化检测试验,结果表明：本文提出的方法能够削弱光照、观测角度等环境差异对建筑物变化检测的影响,显著改善建筑物变化的检测精度。     

Assessment of different topographic correction methods in ALOS AVNIR-2 data over a forest area

Abstract

Because the removal of topographic effects is one the most important pre-processing steps when extracting information from satellite images in digital Earth applications, the problem of differential terrain illumination on satellite imagery has been investigated for at least 20 years. As there is no superior topographic correction method applicable to all areas and all images, a comparison of topographic normalization methods in different regions and images is necessary. In this study, common topographic correction methods were applied on an ALOS AVNIR-2 image of a rugged forest area, and the results were evaluated through different criteria. The results show that the simple correction methods [Cosine, Sun-Canopy-sensor (SCS), and Minnaert correction] are inefficient in exceptionally rough forests. Among the improved correction methods (SCS+C, modified Minnaert, and pixel-based Minnaert), the best result was achieved using a pixel-based Minnaert approach in which a separate correction factor in various slope angles is used. Thus, this method should be considered for topographic correction, especially in forests with severe topography.     

Normalizing shadows in multi-temporal aerial frame imagery using relative radiometric adjustments to support near-real-time change detection

This study addresses the problem of shadows in multi-temporal imagery, which is a key issue with change detection approaches based on image comparison. We apply image-to-image radiometric normalizations including histogram matching (HM), mean-variance (MV) equalization, linear regression based on pseudo-invariant features (PIF-LR), and radiometric control sets (RCS) representing high- and low-reflectance extrema, for the novel purpose of normalizing brightness of transient shadows in high spatial resolution, bi-temporal, aerial frame image sets. Efficient shadow normalization is integral to remote sensing procedures that support disaster response efforts in a near-real-time fashion, including repeat station image (RSI) capture, wireless data transfer, shadow detection (as precursor to shadow normalization), and change detection based on image differencing and visual interpretation. We apply the normalization techniques to imagery of suburban scenes containing shadowed materials of varied spectral reflectance characteristics, whereby intensity (average of red, green, and blue spectral band values) under fully illuminated conditions is known from counterpart reference images (time-1 versus time-2). We evaluate the normalization results using stratified random pixel samples within transient shadows, considering central tendency and variance of differences in intensity relative to the unnormalized images. Overall, MV equalization yielded superior results in our tests, reducing the radiometric effects of shadowing by more than 85 percent. The HM and PIF-LR approaches showed slightly lower performance than MV, while the RCS approach proved unreliable among scenes and among stratified intensity levels. We qualitatively evaluate a shadow normalization based on MV equalization, describing its utility and limitations when applied in change detection. Application of image-to-image radiometric normalization for brightening shadowed areas in multi-temporal imagery in this study proved efficient and effective to support change detection.     

包络线去除的丘陵地区遥感影像阴影信息重建

中国西南丘陵常态山和喀斯特山交错分布,遥感影像普遍存在山体阴影,分布零散且无规律,基于DEM的地形校正模型(C校正等)虽然算法成熟、易于操作,但在复杂地形区存在误差。引入基于相似像元包络线的阴影校正方法(CR校正),按照阴影提取、包络线去除、相似像元寻找和阴影亮度重建的步骤,采用西南丘陵地区Landsat 8 OLI影像进行验证实验。结果表明:CR校正后,阴影区的视觉特征与邻近非阴影区趋于一致,阴影像元亮度有明显提升;校正后影像主要波段标准差减小,与非阴影区参考光谱的相对均方根误差在2.919%以内,最低仅为0.516%;自动分类精度从43.59%提高到61.57%,CR校正有效提高了有阴影的丘陵地区遥感影像质量。     

The effect of atmospheric and topographic correction on pixel-based image composites: Improved forest cover detection in mountain environments

Quantification of forest cover is essential as a tool to stimulate forest management and conservation. Image compositing techniques that sample the most suited pixel from multi-temporal image acquisitions, provide an important tool for forest cover detection as they provide alternatives for missing data due to cloud cover and data discontinuities. At present, however, it is not clear to which extent forest cover detection based on compositing can be improved if the source imagery is firstly corrected for topographic distortions on a pixel-basis. In this study, the results of a pixel compositing algorithm with and without preprocessing topographic correction are compared for a study area covering 9 Landsat footprints in the Romanian Carpathians based on two different classifiers: Maximum Likelihood (ML) and Support Vector Machine (SVM). Results show that classifier selection has a stronger impact on the classification accuracy than topographic correction. Finally, application of the optimal method (SVM classifier with topographic correction) on the Romanian Carpathian Ecoregion between 1985, 1995 and 2010 shows a steady greening due to more afforestation than deforestation.     

一种顾及空间相关性遥感影像辐射度的地形校正算法

地形校正的目的是消除太阳光照对不规则地面地物辐射值的影响。这种影响会使相似植被类型地物的辐射值发生很大的变化。因此,在地形复杂的地区,地形校正是影像预处理的一个重要步骤。传统的基于单像素的地形校正方法,虽然减小了辐射值的变化,但在太阳入射角低的地区常常出现校正过度的情况。针对这种误差进行分析,提出一种考虑了空间相关性的校正算法,并且利用鄂西地区的Landsat7卫星影像进行的试验证明,该算法优于传统的地形校正模型。     

多尺度分割的高分辨率遥感影像变化检测

针对高空间分辨率的遥感影像,提出了一种基于多尺度分割的变化检测算法。采用Mean-Shift分割算法对影像进行多尺度分割,构建了不同尺度上的地理对象,以不同尺度上的地理对象灰度均值构建了变化检测的多尺度特征向量,采用变化矢量分析法获得最后的变化检测结果。以城镇区和农田区的Quick Bird影像对本文算法进行了检验,从精度评价的效果来看,无论城镇区还是农田区,采用面向对象的变化检测方法精度都高于基于单像素的检测方法,且当尺度层数固定时,多尺度组合的变化检测结果优于单一尺度的变化检测结果,对城镇、农田区域的变化检测的精度分别达到87.57%和81.55%。本文算法既可以顾及大面积同质区域变化,又可以反映小的地物目标及边缘部分的变化,能够很好地满足城镇、农田等不同环境背景下的变化检测需求,在国土资源监测中具有一定的应用价值。     

联合像元-深度-对象特征的遥感图像城市变化检测

特征空间的构建和优化对遥感图像识别能力的提高具有重要作用。针对面向对象方法对波段光谱信息利用不足,以及像元识别法无法充分利用图像空间几何等信息的问题,本文建立了新颖的联合像素级和对象级特征的航摄遥感图像城市变化检测方法。首先,充分利用像素级和对象级特征的优势,建立考虑光谱、指数、纹理、几何、表面高度及神经网络深度特征的特征空间;然后,引入LightGBM(light gradient boosting machine)算法对大量特征进行选择研究;最后,采用随机森林识别器对宜兴市2012年和2015年两期遥感图像进行识别,利用变化矩阵进行城市的变化检测。结果表明:联合像元、深度、对象特征和LightGBM特征选择算法的识别效果最好,平均的总体识别精度达到了88.50%,Kappa系数达到0.86,比基于像元、深度或对象特征的识别方法分别提高了10.50%、15.00%和4.00%;城市变化检测精度达到了87.50%。因此,本文方法是利用甚高分辨率航摄遥感图像进行城市变化的检测的有效方法。     

高分辨率遥感影像土地利用变化检测方法研究

提出一种利用高分辨率遥感影像进行土地利用变化检测的方法。以土地利用图为辅助数据,通过土地利用图和遥感影像的配准套合,获取影像像斑;同时,对遥感影像进行基于像素的监督分类,获取概略的类别图;再根据像斑内像素的类别编码完成子像斑的划分。以子像斑为影像分析的基本单位提取特征,以相关系数为相似性测度衡量不同时期子像斑的特征相似性,用ROC曲线(接受者操作特性曲线)代替经验选取的方法自动获取变化阈值,确定像斑是否发生变化。以武汉市区局部QuickBird 2002年和2005年多光谱影像、相同地区2002年1∶10 000土地利用图为实验数据进行了算法的实验,结果显示绝大部分的变化区域都可以被提取出来,实验方法可行。     

Making more out of pixel-level change information: using a neighbourhood approach to improve land change characterization across large and heterogeneous areas

The paper demonstrates two issues; (i) how a ‘moving window approach’, that translates pixel level detected changes to landscape level, can be implemented; (ii) how the approach can overcome the limitations of pixel level change information to characterize change over large areas. First we detected changes from two periods (1986 and 2010) of LULC maps. On the pixel-based changes, we ran focal statistics summation operator separately for selected window sizes (1–10 km). Further, we assessed effect of scale in depicting the pattern and amount of change. The approach is found useful to overcome major shortfalls of pixel-based change characterization. However, varying scale of analysis provide varying amount of change and differently represent change patterns. Thus, implementing the approach over complex and large areas requires multi-scale approach. Subdividing complex and large areas into homogeneous zones can help to implement the multi-scale approach and facilitate the selection of appropriate scale of analysis.     

Impact of Topography on Accuracy of Land Cover Spectral Change Vector Analysis Using AWiFS in Western Himalaya

The present paper discusses the impact of topography on accuracy for land cover classification and “from-to class change using improved spectral change vector analysis suggested by Chen et al. (2003). Two AWiFS sensor images of different dates are used. Double Window Flexible Pace Search (DFPS) is used to estimate threshold of change magnitude for change/no change classes. The topographic corrections show accuracy of 90% (Kappa coefficient 0.7811) for change/no change area as compared to 82% (Kappa coefficient 0.6512) in uncorrected satellite data. Direction cosines of change vector for determining change direction in n-dimensional spectral space is used for image classification with a minimum distance categorizing technique. The results of change detection are compared (i) Improved CVA with conventional two bands CVA and (ii) Improved CVA before and after topographic corrections. The improved CVA with topographic correction consideration using slope match show maximum accuracy of 90% (Kappa coefficient 0.83) as compared to conventional CVA which show maximum accuracy of 82% (Kappa coefficient 0.6624). The overall accuracy of ”from- to class using improved CVA increases from 86% (Kappa coefficient 0.7817) to 90% (Kappa coefficient 0.83) after topographic corrections. The improved CVA with proper topographic corrections is found to be effective for change detection analysis in the rugged Western Himalayan terrain.     

A comparative assessment between object and pixel-based classification approaches for land use/land cover mapping using SPOT 5 imagery

Abstract

Land use/land cover (LULC) classification with high accuracy is necessary, especially in eco-environment research, urban planning, vegetation condition study and soil management. Over the last decade a number of classification algorithms have been developed for the analysis of remotely sensed data. The most notable algorithms are the object-oriented K-Nearest Neighbour (K-NN), Support Vector Machines (SVMs) and the Decision Trees (DTs) amongst many others. In this study, LULC types of Selangor area were analyzed on the basis of the classification results acquired using the pixel-based and object-based image analysis approaches. SPOT 5 satellite images with four spectral bands from 2003 and 2010 were used to carry out the image classification and ground truth data were collected from Google Earth and field trips. In pixel-based image analysis, a supervised classification was performed using the DT classifier. On the other hand, object-oriented (K-NN) image analysis was evaluated using standard nearest neighbour as classifier. Subsequently SVM object-based classification was performed. Five LULC categories were extracted and the results were compared between them. The overall classification accuracies for 2003 and 2010 showed that the object-oriented (K-NN) (90.5% and 91%) performed better results than the pixel-based DT (68.6% and 68.4%) and object-based SVM (80.6% and 78.15%). In general, the object-oriented (K-NN) performed better than both DTs and SVMs. The obtained LULC classification maps can be used to improve various applications such as change detection, urban design, environmental management and zooning.     

A shadow- eliminated vegetation index (SEVI) for removal of self and cast shadow effects on vegetation in rugged terrains

ABSTRACT

The effect of terrain shadow, including the self and cast shadows, is one of the main obstacles for accurate retrieval of vegetation parameters by remote sensing in rugged terrains. A shadow- eliminated vegetation index (SEVI) was developed, which was computed from only red and near-infrared top-of-atmosphere reflectance without other heterogeneous data and topographic correction. After introduction of the conceptual model and feature analysis of conventional wavebands, the SEVI was constructed by ratio vegetation index (RVI), shadow vegetation index (SVI) and adjustment factor (f (Δ)). Then three methods were used to validate the SEVI accuracy in elimination of terrain shadow effects, including relative error analysis, correlation analysis between the cosine of solar incidence angle (cosi) and vegetation indices, and comparison analysis between SEVI and conventional vegetation indices with topographic correction. The validation results based on 532 samples showed that the SEVI relative errors for self and cast shadows were 4.32% and 1.51% respectively. The coefficient of determination between cosi and SEVI was only 0.032 and the coefficient of variation (std/mean) for SEVI was 12.59%. The results indicate that the proposed SEVI effectively eliminated the effect of terrain shadows and achieved similar or better results than conventional vegetation indices with topographic correction.     

Detecting shadows in multi-temporal aerial imagery to support near-real-time change detection

Multi-temporal aerial imagery captured via an approach called repeat station imaging (RSI) facilitates post-hazard assessment of damage to infrastructure. Spectral-radiometric (SR) variations caused by differences in shadowing may inhibit successful change detection based on image differencing. This study evaluates a novel approach to shadow classification based on bi-temporal imagery, which exploits SR change signatures associated with transient shadows. Changes in intensity (brightness from red–green–blue images) and intensity-normalized blue waveband values provide a basis for classifying transient shadows across a range of material types with unique reflectance properties, using thresholds that proved versatile for very different scenes. We derive classification thresholds for persistent shadows based on hue to intensity ratio (H/I) images, by exploiting statistics obtained from transient shadow areas. We assess shadow classification accuracy based on this procedure, and compare it to the more conventional approach of thresholding individual H/I images based on frequency distributions. Our efficient and semi-automated shadow classification procedure shows improved mean accuracy (93.3%) and versatility with different image sets over the conventional approach (84.7%). For proof-of-concept, we demonstrate that overlaying bi-temporal imagery also facilitates normalization of intensity values in transient shadow areas, as part of an integrated procedure to support near-real-time change detection.     

Development of a remotely sensed,historical land-cover change database for rural Chihuahua,Mexico

The purpose of this article is to describe the development of a remotely sensed, historical land-cover change database for the northwestern quarter of Chihuahua, Mexico, The database consists of multi-temporal land-cover classifications and change detection images. The database is developed to facilitate future investigations that examine urban–rural linkages as possible drivers of rural land-use and land-cover changes. To develop the needed land-cover change database, this study uses the North American Landsat Characterization (NALC) MSS triplicates because of their temporal depth and spatial breadth. Challenges exist, however, to effective classification and change detection using the NALC triplicates, including illumination differences across multiple scenes and periods caused by topographic and solar variations and the lack of ground reference data for historic periods. Therefore, creation of the database is a four step process. First, extensive pre-processing is performed to enhance comparability of multi-date images. Pre-processing includes topographic correction, mosaic creation and multi-date radiance normalization. Second, ancillary sources of land-cover data are combined with visual interpretations of enhanced images to define reference pixels used to classify the images using the maximum likelihood algorithm. Third, classification accuracy is assessed. Fourth, post-classification change detection is performed. Results indicate significant image improvements after pre-processing that permit very good overall classification (86.26% classified correctly) and change detection. To conclude, findings are presented that indicate significant changes to arid grasslands/shrublands and forest resources in mountainous regions.     

Integration of hyperspectral and LiDAR data for mapping small water bodies

Inland water bodies are globally threatened by environmental degradation and climate change. On the other hand, new water bodies can be designed during landscape restoration (e.g. after coal mining). Effective management of new water resources requires continuous monitoring; in situ surveys are, however, extremely time-demanding. Remote sensing has been widely used for identifying water bodies. However, the use of optical imagery is constrained by accuracy problems related to the difficulty in distinguishing water features from other surfaces with low albedo, such as tree shadows. This is especially true when mapping water bodies of different sizes. To address these problems, we evaluated the potential of integrating hyperspectral data with LiDAR (hereinafter “integrative approach”). The study area consisted of several spoil heaps containing heterogeneous water bodies with a high variability of shape and size. We utilized object-based classification (Support Vector Machine) based on: (i) hyperspectral data; (ii) LiDAR variables; (iii) integration of both datasets. Besides, we classified hyperspectral data using pixel-based approaches (K-mean, spectral angle mapper). Individual approaches (hyperspectral data, LiDAR data and integrative approach) resulted in 2–22.4 % underestimation of the water surface area (i.e, omission error) and 0.4–1.5 % overestimation (i.e., commission error).The integrative approach yielded an improved discrimination of open water surface compared to other approaches (omission error of 2 % and commission error of 0.4 %). We also evaluated the success of detecting individual ponds; the integrative approach was the only one capable of detecting the water bodies with both omission and commission errors below 10 %. Finally, the assessment of misclassification reasons showed a successful elimination of shadows in the integrative approach. Our findings demonstrate that the integration of hyperspectral and LiDAR data can greatly improve the identification of small water bodies and can be applied in practice to support mapping of restoration process.     

Evaluation of different topographic correction methods for Landsat imagery

The recent free availability of Landsat historical data provides new potentials for land-cover change studies. Multi-temporal studies require a previous radiometric and geometric homogenization of input images, to better identify true changes. Topographic normalization is one of the key steps to create consistent and radiometricly stable multi-temporal time series, since terrain shadows change throughout time. This paper aims to evaluate different methods for topographic correction of Landsat TM-ETM+ data. They were assessed for 15 ETM+ images taken under different illumination conditions, using two criteria: (a) reduction of the standard deviation (SD) for different land-covers and (b) increase in temporal stability of a time series for individual pixels. We observed that results improve when land-cover classes where processed independently when applying the more advanced correction algorithms such as the C-correction and the Minnaert correction. Best results were obtaining for the C-correction and the empiric–statistic correction. Decreases of the SD for bare soil pixels were larger than 100% for the C-correction and the empiric–statistic correction method compared to the other correction methods in the visible spectrum and larger than 50% in the IR region. In almost all tests the empiric–statistic method provided better results than the C-correction. When analyzing the multi-temporal stability, pixels under bad illumination conditions (northern orientation) improved after correction, while a deterioration was observed for pixels under good illumination conditions (southern orientation). Taken this observation into account, a simple but robust method for topographic correction of Landsat imagery is proposed.