深度学习U-Net方法及其在高分辨卫星影像分类中的应用

高分辨率遥感影像有精确的几何结构和空间布局,但是光谱信息有限,增大了对光谱特征相似地物的分类难度。针对高分辨率遥感影像分类的问题,采用深度学习U-Net模型分类方法。基于黑河下游额济纳绿洲高分二号遥感影像,通过U-Net模型提取胡杨、柽柳、耕地、草地和裸地五种地物覆被类型,分类总体精度和Kappa系数分别为85.024%和0.795 6,并与传统的支持向量机（SVM, Support Vector Machine）和面向对象的分类方法比较,结果表明：相对于SVM和面向对象,基于U-Net模型的高分辨率卫星影像地物覆被分类,能够更好地对地物本质特征进行提取,分类效果较好,满足精度要求。     

基于面向对象分类技术的景观信息提取研究

依据高分辨率遥感影像的特点,如何充分地利用影像的光谱信息和空间信息以及地学特征进行更为微观的遥感监测或大比例尺制图是高分辨率遥感研究的重要内容之一。本文以地形复杂的武夷山自然保护区为研究区域,以SPOT5原始影像为数据源,采用面向对象的多尺度分割方法,实现了不同尺度地物信息的分层提取。基于上层的分类结果对特定地物进行影像分割,选择合适的特征参数,并通过多次试验建立影像对象的隶属度函数,或利用最邻近分类法,逐级分层地提取了研究区的景观信息。研究结果显示：利用面向对象分类方法可以快速方便地对地形复杂条件下研究区的SPOT5遥感影像的景观信息提取,精度为76%,为高分辨率遥感影像的信息提取提供了更为快速、有效的技术途径。     

基于多时相Sentinel-2影像的黑河中游玉米种植面积提取研究

玉米是黑河中游种植面积最大的农作物,生长期需水量大、蒸散量高。准确获取玉米种植面积对该区域农作物种植结构调整、水资源合理规划有重要参考意义。基于2019年4月至9月Sentinel-2多时相影像,采用随机森林算法开展了黑河中游玉米种植面积提取研究。研究方法分为两类—直接提取法和两步提取法。进一步探讨了多时间信息量对玉米种植面积提取精度的影响以及各输入特征参数在玉米面积提取过程中的重要性表现。结果表明：基于Sentinel-2多时相影像,直接提取法和两步提取法均可高精度地提取研究区玉米种植面积,特别是两步提取法,玉米分类总体精度可达85.03%,F1_Score为0.70,Kappa系数为0.83;与单幅影像相比,多时相影像可获取不同作物的物候信息,有效减少作物错分/漏分,提高作物分类精度。该方法对基于高分辨率光学影像结合机器学习方法获取具有高度异质性的作物信息具有重要的参考价值。     

应用剖分遥感影像模板快速提取作物种植面积

全球剖分理论为全球海量遥感影像数据的组织管理和多尺度遥感影像的作物提取和识别提供新的解决思路．结合基于地图分幅扩展的全球剖分模型及其剖分面片的几何特征,阐述剖分遥感影像模板的概念模型和数据模型,提出利用剖分遥感影像模板来提取作物种植面积的处理流程．并给出不同尺度范围提取作物面积适宜选取的剖分级别和影像分辨率。采用高分辨率遥感影像初步尝试对河南省许昌地区小麦种植面积进行提取,通用遥感影像处理软件相比,其精度和速度都有一定的提高．     

多时相影像的冬小麦种植面积提取及估产

针对多时相影像的农作物种植面积难以实现统一精确提取、不能高效地进行遥感估产研究的问题,以河南省濮阳市为研究区,基于Landsat TM影像,采用基于伪不变特征的相对辐射校正方法,在深入分析濮阳市内6类典型地物光谱的基础上,构建决策树提取冬小麦种植面积。然后,基于MODIS植被指数产品,结合相应年份统计数据进行植被指数校正,分别利用校正后关键生育期的归一化植被指数累计值和增强型植被指数累计值与冬小麦产量进行回归分析,建立冬小麦产量预测模型,利用2011年的产量进行验证。结果表明:各年份冬小麦的提取面积精度均在96.3%以上,利用归一化植被指数和增强型植被指数构建的估产模型,R2分别为0.834和0.926,估产精度分别为95.36%和96.44%。该研究可为市域冬小麦种植区的统一高效提取以及冬小麦产量预测提供参考。     

地物大小、对象尺度、影像分辨率的关系分析

遥感数据的分辨率越来越高, 给地物信息提取提出了新的挑战。利用基于像元的分类技术和基于多尺度分割的面向对象分类技术对高分辨率影像进行分类实验, 分析地物大小、对象尺度与影像分辨率的关系。实验结果表明不同地物由于其空间尺度不同, 与之相适宜的空间分辨率和对象尺度也不同, 在适宜分辨率的影像提取有较高的精度, 在适宜的对象尺度上提取对象信息有更高的精度。分析也表明面向对象的多尺度影像分类技术适应了不同地物有其相适宜的空间分辨率, 在适宜尺度影像层中提取地物, 其分类精度大大高于基于像元的分类方法。     

基于地理加权的k-|NN高分辨率遥感影像分类算法改进

与中低分辨率相比,高分辨率遥感影像的信息比较丰富,在使用常规k-NN分类方法基于像元进行高分辨率遥感影像分类时会产生大量的“椒盐噪声”和地物类别错分。根据地理学第一定律,引入地统计模型,将地理权重加入到常规k-NN分类方法中,形成新的地理权重k-NN分类器(Geographically Weighted k-NN,GWk-NN)。该方法首先通过条件概率函数计算出训练样本数据的空间分布特征,然后通过地统计模型对空间分布特征进行拟合,为每种地物选择合适的权重模型,这样既保留了遥感影像中地物的光谱特征,又融入了地物的空间特征,在一定程度上减少甚至消除了“椒盐噪声”,提高了分类精度。GWk\|NN和常规k\|NN分类器分析对比表明：GWk-NN分类方法提高了高分辨率影像的分类精度。     

结合地籍数据的高密度城区面向对象遥感分类 

利用高分辨率遥感影像和GIS辅助数据,对高密度城区进行面向对象的土地利用覆被分类研究。使用NAIP高分辨率航空遥感影像,在多尺度影像分割的基础上,针对特定地物选择合适的影像分割参数。采用决策树方法建立高密度城市地区的分类规则,并结合该地区地籍图数据作为辅助数据,逐步进行高密度城市地区地物信息提取。利用辅助数据进行面向对象的遥感分类效果优于单纯依靠遥感影像进行的分类,且有效提取了道路和复杂的房屋等信息,得到了理想的分类结果,其总分类精度从常规面向对象方法的84.08%提高到89.79%。利用辅助数据进行遥感分类提高了高分辨率遥感影像的分类精度,说明了利用辅助数据进行遥感分类方法的有效性。     

WorldView卫星16波段影像光谱特征分类

传统的高分辨率遥感卫星光谱分辨率较低,WorldView卫星在8个可见光G近红外多光谱波段的基础上,新增加的８个短波红外(short wave infrared,SWIR)影像,有助于提高影像提取地物信息能力。分析了WorldView卫星的16波段影像上各种地物的光谱特征和分类性能,提出了新的植被指数、水体指数和建成区指数。实验表明,相比于8波段影像,使用16波段影像分类能够显著提高各类地物特别是裸地、建筑物和道路的分类精度,总体精度提高约5.5％。基于16波段设计的新地物特征指数能更好地避免干扰地物,通过简单阈值提取地物,取得较高的提取精度。     

WorldView卫星16波段影像光谱特征分类

传统的高分辨率遥感卫星光谱分辨率较低,WorldView卫星在8个可见光-近红外多光谱波段的基础上,新增加的8个短波红外(short wave infrared,SWIR)影像,有助于提高影像提取地物信息能力。分析了WorldView卫星的16波段影像上各种地物的光谱特征和分类性能,提出了新的植被指数、水体指数和建成区指数。实验表明,相比于8波段影像,使用16波段影像分类能够显著提高各类地物特别是裸地、建筑物和道路的分类精度,总体精度提高约5.5%。基于16波段设计的新地物特征指数能更好地避免干扰地物,通过简单阈值提取地物,取得较高的提取精度。     

Object-based land cover classification of shaded areas in high spatial resolution imagery of urban areas: A comparison study

A significant proportion of high spatial resolution imagery in urban areas can be affected by shadows. Considerable research has been conducted to investigate shadow detection and removal in remotely sensed imagery. Few studies, however, have evaluated how applications of these shadow detection and restoration methods can help eliminate the shadow problem in land cover classification of high spatial resolution images in urban settings. This paper presents a comparison study of three methods for land cover classification of shaded areas from high spatial resolution imagery in an urban environment. Method 1 combines spectral information in shaded areas with spatial information for shadow classification. Method 2 applies a shadow restoration technique, the linear-correlation correction method to create a “shadow-free” image before the classification. Method 3 uses multisource data fusion to aid in classification of shadows. The results indicated that Method 3 achieved the best accuracy, with overall accuracy of 88%. It provides a significantly better means for shadow classification than the other two methods. The overall accuracy for Method 1 was 81.5%, slightly but not significantly higher than the 80.5% from Method 2. All of the three methods applied an object-based classification procedure, which was critical as it provides an effective way to address the problems of radiometric difference and spatial misregistration associated with multisource data fusion (Method 3), and to incorporate thematic spatial information (Method 1).     

结合像元形状特征分割的高分辨率影像面向对象分类

针对高分辨率遥感影像空间分辨率高,结构形状、纹理、细节信息丰富等特点,提出一种新的融合特征的面向对象影像分类方法来提取城市空间信息。基本过程包含以下4个方面:①提取影像的几何纹理等结构;②融合几何与纹理特征的面向对象影像分割;③提取对象的形状、纹理和光谱特征,并优选最佳特征子集;④最后基于支持向量机(SVM)完成面向对象的影像分类。通过对福州IKONOS影像数据实验,结果表明融入影像特征后的分割效果明显优于原始影像的分割结果,而信息最大化(mRMR)的特征选择能够快速地获得较好的特征子集。通过与eCognition最邻近分类方法比较,表明本文方法的分类总体精度大约提高了6%,效果显著。     

联合快舟一号影像纹理信息的城市土地覆盖分类

仅依靠光谱信息无法满足高分辨率遥感分类的应用需求,辅之以纹理特征信息进行分类,可提高影像分类精度。利用KZ\|1卫星影像和Landsat\|8卫星影像数据,基于面向对象的影像分割法和灰度共生矩阵纹理分析法对新疆石河子市局部城区进行了地表覆盖分类实验,将不同空间分辨率的全色影像纹理信息、光谱信息构成多种影像特征组合进行分类比较研究,以选择最佳的分类特征集。结果表明:KZ-1影像能为城市区域的土地覆盖分类提供丰富的纹理信息,面向对象的影像分割可较好地利用高分辨率数据的几何结构信息实现优化的影像分割,从而提高多光谱影像的分类精度,总体分类精度为90.06%,Kappa系数为87.93%,比单纯利用光谱信息分类的总体精度提高了8.02%,Kappa系数提高了9.65%,表明KZ\|1数据可为光谱分类提供丰富的纹理信息,从而提高城市区域的土地覆盖分类精度。     

Effect of coregistration error on patchy target detection using high-resolution imagery

Many factors influence classification accuracy and a typical error budget includes uncertainty arising from the 1) selection of processing algorithms, 2) selection of training sites, 3) quality of orthorectification, and 4) atmospheric effects. With the development of high spatial resolution imagery, the impact of errors in geographic coregistration between imagery and field sites has become apparent - and potentially limiting - for classification applications, especially those involving patchy target detection. The goal of this study was to document and quantify the effect of coregistration error between imagery and field sites on classification accuracy. Artificial patchy targets were randomly placed over a study area covered by a QuickBird image. Classification accuracy of these targets was assessed at two levels of coregistration. Results showed that producer's accuracy of target classification increased from 37.5% to 100% between low and high levels of coregistration respectively. In addition, “Error due to Location”, a measure of how well pixels were located within respective classes, decreased to zero at high coregistration levels. This study highlights the importance of considering coregistration between imagery and field sites in the error budget, especially with studies involving high spatial resolution imagery and patchy target detection.     

LiDAR点云支持下地物精细分类的实现方法

在遥感数据分类中,获取精细的地物类别无疑能够传递更加丰富的信息量,进一步加深对遥感数据的理解和解译。在机载LiDAR点云高程数据的支持下,提出并实现了遥感影像上地物精细分类的方法。为保证高精度地同种地物再划分,综合考虑配准、辅助数据源、首次回波、点云密度及影像空间分辨率4种因素,并重点解决了点云密度与影像空间分辨率不匹配的问题,利用决策树显著地提高了影像上建筑物、植被的分类数量,使点云与影像联合分类的优势得到体现,达到了分类精度与地物类别数量相统一的目的。     

Mapping reedbed habitats using texture-based classification of QuickBird imagery

Many organisms rely on reedbed habitats for their existence, yet, over the past century there has been a drastic reduction in the area and quality of reedbeds in the UK due to intensified human activities. In order to develop management plans for conserving and expanding this threatened habitat, accurate up-to-date information is needed concerning its current distribution and status. This information is difficult to collect using field surveys because reedbeds exist as small patches that are sparsely distributed across landscapes. Hence, this study was undertaken to develop a methodology for accurately mapping reedbeds using very high resolution QuickBird satellite imagery. The objectives were to determine the optimum combination of textural and spectral measures for mapping reedbeds; to investigate the effect of the spatial resolution of the input data upon classification accuracy; to determine whether the maximum likelihood classifier (MLC) or artificial neural network (ANN) analysis produced the most accurate classification; and to investigate the potential of refining the reedbed classification using slope suitability filters produced from digital terrain data. The results indicate an increase in the accuracy of reedbed delineations when grey-level co-occurrence textural measures were combined with the spectral bands. The most effective combination of texture measures were entropy and angular second moment. Optimal reedbed and overall classification accuracies were achieved using a combination of pansharpened multispectral and texture images that had been spatially degraded from 0.6 to 4.8 m. Using the 4.8 m data set, the MLC produced higher classification accuracy for reedbeds than the ANN analysis. The application of slope suitability filters increased the classification accuracy of reedbeds from 71% to 79%. Hence, this study has demonstrated that it is possible to use high resolution multispectral satellite imagery to derive accurate maps of reedbeds through appropriate analysis of image texture, judicious selection of input bands, spatial resolution and classification algorithm and post-classification refinement using terrain data.     

Using QuickBird imagery and a production efficiency model to improve crop yield estimation in the semi-arid hilly Loess Plateau,China

Crop yield is a key element in rural development and an indicator of national food security. A method that could estimate crop yield over large hilly areas would be highly desirable. Methods including high spatial resolution satellite imagery have the potential to achieve this objective. This paper describes a method of integrating QuickBird imagery with a production efficiency model (PEM) to estimate crop yield in Zhonglianchuan, a hilly area on Loess Plateau, China. In the PEM model, crop yield is a function of the photosynthetic active radiation (PAR), fraction of absorbed photosynthetically active radiation (f_APAR) and light-use efficiency (LUE). Based on the high spatial resolution QuickBird imagery, a land cover classification is used to attribute a class-specific LUE. The f_APAR is related to spectral vegetation indices (SVI), which can be derived from the satellite images. The LUE, f_APAR and incident PAR data were combined to estimate the crop yield. Farmer-reported crop yield data in 80 representative plots were used to validate the model output. The results indicated QuickBird imagery can improve the accuracy of predicted results relative to the Landsat TM image. The predicted yield approximated well with the data reported by the farmers (r² = 0.86; n = 80). The spatial distributions of crop yield derived here also offers valuable information to manage agricultural production and understand ecosystem functioning.     

Object‐based land‐cover classification for the Phoenix metropolitan area: optimization vs. transportability

An object‐based approach was utilized in the development of two urban land‐cover classification schemes on high‐resolution (0.6 m), true‐colour aerial photography of the Phoenix metropolitan area, USA. An initial classification scheme was heavily weighted by standard nearest‐neighbour (SNN) functions generated by samples from each of the classes, which produced an enhanced accuracy (84%). A second classification was developed from the initial classification scheme in which SNN functions were transformed into a fuzzy‐rule set, creating a product transportable to different areas of the same imagery, or for land‐cover change detection with similar imagery. A comprehensive accuracy assessment revealed a slightly lower overall accuracy (79%) for the rule‐based classification. We conclude that the transportable classification scheme is satisfactory for general land‐cover analyses; yet classification accuracy can be enhanced at site‐specific venues with the incorporation of nearest‐neighbour functions using class samples.     

Incorporating texture into classification of forest species composition from airborne multispectral images

Although research with digital airborne remote sensing data has been undertaken in different ecoregions to classify forested areas, the potential role of such imagery in deriving information to assist forest management has not yet been fully defined. The objective of this study was to determine the extent that the addition of texture could improve spectral classification of high spatial resolution images (pixel size 1m). These images represented pure and mixed wood forest stands from ecoregions in Alberta and New Brunswick, Canada. This study employed a judicious, selective application of texture to stands within a hierarchical classification framework. In Alberta, the addition of texture made a modest improvement in classification accuracy from 60% to 65%. In New Brunswick, the application of texture to selected land cover types resulted in an overall 12% improvement in classification accuracy. The addition of image texture increased classification accuracy for high spatial detail imagery relative to low spatial detail imagery. Incorporating texture into classification also improved classification accuracies for hardwood stands more so than for softwood stands, but greater attention to stand structure and composition will be needed in future work. Classification accuracies on the order of 60-65% were achieved with simple texture derivatives, maximum likelihood decision rules and conventional classification methods.     

Towards a set of agrosystem-specific cropland mapping methods to address the global cropland diversity

Accurate cropland information is of paramount importance for crop monitoring. This study compares five existing cropland mapping methodologies over five contrasting Joint Experiment for Crop Assessment and Monitoring (JECAM) sites of medium to large average field size using the time series of 7-day 250 m Moderate Resolution Imaging Spectroradiometer (MODIS) mean composites (red and near-infrared channels). Different strategies were devised to assess the accuracy of the classification methods: confusion matrices and derived accuracy indicators with and without equalizing class proportions, assessing the pairwise difference error rates and accounting for the spatial resolution bias. The robustness of the accuracy with respect to a reduction of the quantity of calibration data available was also assessed by a bootstrap approach in which the amount of training data was systematically reduced. Methods reached overall accuracies ranging from 85% to 95%, which demonstrates the ability of 250 m imagery to resolve fields down to 20 ha. Despite significantly different error rates, the site effect was found to persistently dominate the method effect. This was confirmed even after removing the share of the classification due to the spatial resolution of the satellite data (from 10% to 30%). This underlines the effect of other agrosystems characteristics such as cloudiness, crop diversity, and calendar on the ability to perform accurately. All methods have potential for large area cropland mapping as they provided accurate results with 20% of the calibration data, e.g. 2% of the study area in Ukraine. To better address the global cropland diversity, results advocate movement towards a set of cropland classification methods that could be applied regionally according to their respective performance in specific landscapes.