Pattern decomposition method in the albedo space for Landsat TM and MSS data analysis

We have developed a 'pattern decomposition method' based on linear spectral mixing of ground objects for n-dimensional satellite data. In this method, spectral response patterns for each pixel of an image are decomposed into three components using three standard spectral shape patterns determined from the image data. Applying this method to Landsat Thematic Mapper data, six-dimensional data are successfully transformed into three-dimensional data. Nearly 94% of the information in the six-dimensional data is retained in the three components. This method is very useful for classifying and monitoring changes in land cover.     

A new vegetation index derived from the pattern decomposition method applied to Landsat‐7/ETM+ images in Mongolia

The goal of this study was to estimate vegetation coverage and map the land‐cover in an experimental field (60×60 km) near Mandalgobi, Mongolia using Landsat‐7/ETM+ data for ground truthing in the Advanced Earth Observing Satellite II (ADEOS‐II) Mongolian Plateau Experiment (AMPEX). We measured soil moisture, vegetation coverage, and vegetation moisture in the field at 49 grid points around the time that the Aqua satellite passed over the area. We also surveyed the land‐cover in the field. Using ground‐based data and characteristics of spectral reflectance, we attempted to extract vegetation information from satellite data using the pattern decomposition method, which is a type of spectral mixture analysis. This method uses normalized spectral shapes as endmembers, which do not change between scenes. We defined an index using the pattern decomposition coefficients to analyse sparsely vegetated areas. The index showed a linear relationship with vegetation coverage. The vegetation coverage was estimated for the study site, and the average coverage at the site was 21.4%. Land‐cover types were classified using the index and the pattern decomposition coefficients; the kappa coefficient was 0.75. The index was useful for estimating vegetation coverage and land‐cover mapping for semiarid areas.     

Sensor‐independent analysis method for hyperspectral data based on the pattern decomposition method

This paper describes a modified pattern decomposition method with a supplementary pattern. The proposed approach can be regarded either as a type of spectral mixing analysis or as a kind of multivariate analysis; the later explanation is more suitable considering the presence of the additional supplementary patterns. The sensor‐independent method developed herein uses the same normalized spectral patterns for any sensor: fixed multi‐band (1260 bands) spectra serve as the universal standard spectral patterns. The resulting pattern decomposition coefficients showed sensor independence. That is, regardless of sensor, the three coefficients had nearly the same values for the same samples. The estimation errors for pattern decomposition coefficients depended on the sensor used. The estimation errors for Landsat/MSS and ALOS/AVNIR‐2 were larger than those of Landsat/TM (ETM+), Terra/MODIS and ADEOS‐II/GLI. The latter three sensors had negligibly small errors.     

An evaluation of the effect of the spectral response function of satellite sensors on the precision of the universal pattern decomposition method

In previous studies of the universal pattern decomposition method (UPDM), the band width has been used to calculate standard spectral pattern vectors, without consideration of the effect of spectral response functions (SRFs). This study revised the UPDM to further reduce sensor dependence, by taking into account the effect of SRFs. Both the UPDM and the revised UPDM (RUPDM) were applied to MODIS and ETM+?data acquired over the Three Gorges region of China. The reconstruction accuracy was significantly greater when the RUPDM was used, with a relative decrease in the mean χ² of more than 14%. Using the new method, the dependence of the decomposition coefficients and vegetation index (VIUPD) on the sensor also decreased, with their linear regression factors approximately equal to one. These increases in accuracy indicate that the RUPDM further reduces sensor dependence and hence can outperform the UPDM in data retrieval.     

航空影像辅助下的城区机载LiDAR汽车目标检测方法

机载激光雷达(LiDAR)技术的出现为地面汽车目标检测提供了新的途径。为了从机载LiDAR点云数据中提取汽车对象,根据不同地物的属性特征,提出了一种航空影像辅助下的城区机载LiDAR汽车目标检测方法。首先利用形态学开重建滤波完成地面和地物的分类,然后在地物点的基础上结合正射影像,通过归一化植被指数(NDVI)特征完成对植被和非植被地物的初步分类,最后在非植被地物的基础上,根据地物对象的形状特征及高程信息完成汽车和建筑物及阴影植被等非汽车对象的分类,从而完成汽车目标的提取工作。3个实验区的计算结果表明:该方法能有效从LiDAR点云中提取汽车目标,正确度和完整度的均值分别为95%和85%,满足实用性要求。     

TM digital processing of a tropical forest region in southeastern Mexico

A humid forest in the neotropical area of Los Tuxtlas, in southeastern Mexico has been used as a test area (900km²) for classification of landscape and vegetation by means of Landsat Thematic Mapper (TM) data, aerial photography and 103 ground samples. The area presents altitudinal variations from sea level to 1640m, providing a wide variety of vegetation types. A hybrid (supervised/unsupervised) classification approach was used, defining spectral signatures for 14 clustering areas with data from the reflective bands of the TM. The selected clustering areas ranged from vegetation of the highlands and the rain forest to grassland, barren soil, crops and secondary vegetation. The digital classification compared favourably with results from aerial photography and with those from a multivariate analysis of the 103 ground data. The statistical evaluation (error matrix) of the classified image indicated an overall 84·4 per cent accuracy with a kappa coefficient of agreement of 0·83. A geographical information system (GIS) was used to compile a land unit and a vegetation map. The TM data allowed for delineation of boundaries in the land unit map, and for a finer differentiation of vegetation types than those identified during field work. Digital value patterns of several information classes are shown and discussed as an indirect guide of the spectral behaviour of vegetation of highlands, rain forest, secondary vegetation and crops. The method is considered applicable to the inventory of other forested areas, especially those with significant variations in vegetation.     

Comparison of the sensor dependence of vegetation indices based on Hyperion and CHRIS hyperspectral data

In previous studies of the universal pattern decomposition method (UPDM), spectral shifts, which are very common in hyperspectral imaging spectrometers, were not taken into account when calculating standard spectral pattern vectors. This study evaluated the effect of spectral shifts on the sensor dependence of the vegetation index based on the UPDM (VIUPD) and 11 other vegetation indices (VIs). Spectral shifts were calculated using Gao's spectrum-matching method. The influences of smoothing techniques (moving average and Savitzky–Golay filters) on the consistency of these VIs were also evaluated and compared. Data from the typical narrowband imaging spectrometers, Hyperion and the Compact High Resolution Imaging Spectrometer (CHRIS), were chosen for the study. For all VIs, both smoothing and spectral calibration changed the consistency between Hyperion and CHRIS. Spectral calibration had a positive effect on the majority of VIs, whereas smoothing improved the performance of some VIs but decreased the consistency of others. When compared with spectral calibration and Savitzky–Golay smoothing, moving average generated greater variations within the results. Among the smoothing techniques employed, moving average smoothing exhibited a larger distortion of VI sensor dependency than that of Savitzky–Golay smoothing of the same order. VIUPD based on narrowband hyperspectral data was sensitive to spectral operations (spectral calibration and smoothing). For VIUPD, spectral calibration increased its sensor independence, whereas smoothing had a negative effect. After spectral calibration, VIUPD was more sensor independent than any other VI examined in this study.     

一种应用机载LiDAR数据和高分辨率遥感影像提取城市绿地信息的方法

提出了一种新的面向对象的城市绿地信息两阶段提取方法。该方法分阶段使用高分辨率遥感影像的光谱和2维形态信息以及机载LiDAR数据的3维形态信息作为分类依据。第1阶段,影像首先被分割为对象,对象被分类为无阴影的植被、阴影下的植被、水体、建筑物、空地和阴影6类地物;无阴影的植被和阴影下的植被合并为城市绿地对象,在第2阶段,将LiDAR数据产生的归一化数字表面模型nDSM与绿地对象叠加,计算每个对象的3维形态属性,进一步将绿地对象细分为草坪、灌木和乔木。以美国休斯敦中心城区为例,介绍了方法流程。精度分析表明,绿地的分类精度达到9346%;方法中的主要误差来源于遥感影像当中的建筑物阴影以及生成数字地形模型时所产生的误差。     

Classification of lidar bare-earth points,buildings, vegetation,and small objects based on region growing and angular classifier

In recent years, light detection and ranging (lidar) systems have been intensively used in different urban applications such as map updating, communication analysis, virtual city modelling, risk assessment, and monitoring. A prerequisite to enhance lidar data content is to differentiate ground (bare earth) points that yield digital terrain models and off-terrain points in order to classify urban objects and vegetation. The increasing demand for a fast and efficient algorithm to extract three-dimensional urban features was the motive behind this work. A new combined approach to extract bare-earth points is proposed, and a novel methodology to automatically classify airborne laser data into different objects in an urban area is presented. In addition, a new concept of angular classification is introduced to differentiate buildings from vegetation and other small objects. The new angular classifier analyses the distribution of bare-earth points around unclassified point clusters to determine whether a cluster can be classified either as building or as vegetation. The experimental results confirm the high accuracy achieved to automatically classify urban objects in flat complex areas.     

The area extraction of winter wheat in mixed planting area based on Sentinel-2 a remote sensing satellite images

ABSTRACT

The traditional area extraction method mainly depends on manual field survey methods, it is workload, slow and high cost. While remote sensing technology has the advantages of accuracy, rapidity, macroscopic and dynamic, which has become an effective means to extract crop growing area. In this paper, we took Kaifeng City in Henan Province as the study area. Firstly, we explored the advantages of Sentinel-2A RENDVI in crop identification. Then used the supervised classification SVM, object-oriented classification method and assisted with field measured data to extract the winter wheat planting area, the characteristics of the two methods were compared and analysed. Finally, we combined the above two classification methods and proposed a new classification method V2OAE to remove unnecessary influencing factors. The experiment results showed that RENDVI has better recognition ability than the NDVI (Normalized Difference Vegetation Index) in distinguishing vegetation with similar spectrum, the classification effect of object-oriented classification is better than supervised classification SVM, and our classification method removes unnecessary influence factors in the results of object-oriented classification, which is further improve the monitoring accuracy.

Firstly, we have preprocessed the Sentinel-2A image data, its steps are: (1) In the first step, we made radiation calibration for remote sensing images to eliminate the image distortion caused by external factors, data acquisition and transmission systems and so on; (2) In the second step, we made atmospheric correction to eliminate changes in the spectral feature of remote sensing images caused by atmospheric absorption or scattering; (3) In the third step, we made band resampling to unify the resolution of remote sensing images and facilitate the mathematical combination operation of vegetation index; (4) In the fourth step, we made mosaic and cutting to get preprocessed remote sensing images of Kaifeng City. Secondly, we analysed the spectral features of each object and established the interpretation mark with the field measured data. then we explored the ability to identify the ground objects based on NDVI(Normalized Difference Vegetation Index) and RENDVI. Third, we used the rule-based object-oriented classification method and SVM classification to extract the planting area of the study area, the input definition of SVM is spectral feature images of ground objects and the output definition of SVM is the recognition result of ground objects in the process of data training. Then the advantages and disadvantages of the two methods in classification results were analysed. Finally, In order to extract winter wheat information more accurately, we combined the above two classification methods and proposed a new classification method V2OAE (Vector Object Oriented Area Extraction) to remove unnecessary influencing factors, then the winter wheat planting area in Kaifeng City was statistically obtained.     

An Improved Method for Object Extracting based on Context Information Using Multisouce Data

This paper proposed a new method which combines the airborne LiDAR data with aerial image to extract Rolling Stones on mountainous.Firstly,the aerial image is processed with multi-scale segmentation to get segmentation objects,and the LiDAR data are processed by classification,interpolation,difference for elevation information.Then compute the segmentation object based on visible-band difference vegetation index to remove the interference of vegetation information,and the nonvegetated segmentation objects are obtained.In order to effectively use the shadow,this paper put forward the normalized difference shadow index and use threshold segmentation to get shadow object.And then the automatic extraction algorithm based on the shadow and elevation information is used to preliminary obtain the rolling stones information.Finally,The height threshold filtering is set according to the actual demand to get the final rolling information.This paper took a certain area of Hong Kong aviation image and LiDAR data as experimental data to validate the proposed method.The results show that the method can well extract the Rolling Stones and effectivly distinguish the exposed bedrock,roads and similar spectral information of ground objects as Rolling Stones.The extraction accuracy of Rolling Stones is above 88% which basically satisfies the needs of rockfall in lands department.     

Salt marsh vegetation radiometry: Data analysis and scaling

This paper aims to determine the optimal procedure for classifying salt marsh vegetation from hyperspectral data and to establish relationships between airborne and ground measurement properties. The study is carried out on data collected in the Lagoon of Venice (Italy). Spectral angle mapping proves to be a reliable classification procedure, and spectral differentiation is seen to improve separability of vegetation types. Further, scaling relationships are derived to link the value of the variance of data aggregated at different scales, allowing the determination of data variability at coarse resolution on the basis of ground measurements. The comparison between the theoretical, up-scaled, values of standard deviation and those computed from remote sensing data shows a good agreement supporting the derived relations. Finally, a scaling relationship is established for spectral angles, which may be useful in determining an optimal threshold angle from ground data.     

Detecting wind disturbance severity and canopy heterogeneity in boreal forest by coupling high-spatial resolution satellite imagery and field data

Wind disturbance events can impact spatially heterogeneous patterns in vegetation structure and disturbance severity in forested landscapes. Characterizing these patterns in forested ecosystems with remote sensing data has been a persistent challenge as variation in severity may be heterogeneous at fine spatial scales. Yet the degree and pattern of disturbance severity are an important influence on successional dynamics. This study explored how spectral and textural characteristics of high-spatial resolution IKONOS imagery reflected patterns of disturbance severity across a windstorm damaged, 121-km² area of the Boundary Waters Canoe Area Wilderness (BWCAW) in northeastern Minnesota, USA. In this study, spectral and spatial features of high-spatial resolution (1-m panchromatic and 4-m multispectral) IKONOS satellite imagery from a single post-disturbance date are coupled with field observations of disturbance within 0.045-ha field plots to access the potential for empirically modeling disturbance severity across this heterogeneous landscape within the BWCAW. Combining textural and spectral features led to a multiple regression model that explained 68% of the variance, and predicted disturbance severity equally well for ground data not included in the model development. The results suggest the utility of combining spatial and spectral data for detecting differences in forest structure caused by ecological processes such as disturbance.     

Biophysical modelling and NDVI time series to project near‐term forage supply: spectral analysis aided by wavelet denoising and ARIMA modelling

Point‐based biophysical simulation of forage production coupled with 1‐km AVHRR NDVI data was used to determine the feasibility of projecting forage conditions 84 days into the future to support stocking decision making for livestock production using autoregressive integrated moving average (ARIMA) with Box and Jenkins methodology. The study was conducted at three highly contrasting ecosystems in South Texas over the period 1989–2000. Wavelet transform was introduced as a mathematical tool to denoise the NDVI time series. The simulated forage production, NDVI and denoised NDVI (DeNDVI) were subject to spectral decomposition for the detection of periodicities. Spectral analysis revealed bimodal vegetation growth patterns in Southwestern Texas. A yearly cycle (364 days) of peak vegetation production was detected for the three study sites, another peak forage production was revealed by spectral analysis at 182 days following the first peak in vegetation production. A similar trend was found for the NDVI imageries sensing the study sites. Wavelet denoising of NDVI signal was effective in revealing clear periodicities in one study site where maximum variability of NDVI was noted.

The Box and Jenkins ARIMA modelling approach was used as a forecasting method for near‐term forage production to assist range managers in proactive operational stocking decisions to mitigate drought risk. Using denoised NDVI provided forage projections with the lowest standard error prediction (SEP) throughout the forecast 84‐day periods. However, acceptable SEP was only achieved up to 6 weeks into a projection for the forage‐only based forecasts. The ARIMA forecasting methodology appears to offer a new approach to help managers of livestock production through the creation of near real‐time early warning systems. Using satellite‐derived NDVI data as a covariate improved the forecast quality and reduced the standard error of forecast in three highly contrasting sites. Denoising the NDVI data using wavelet methods further improved the forecast quality in all study sites.

The integration of AVHRR NDVI data and biophysical simulation of forage production appears a promising approach for assisting decision makers in a positive manner by assessing forage conditions in response to emerging weather conditions and near real‐time projection of available forage for grazing animals.     

马尔可夫随机场的空间相关模型在非负矩阵分解线性解混中的应用

针对基于非负矩阵分解（NMF）的高光谱解混存在的初始化与"局部极小"等问题,提出一种基于马尔可夫随机场（MRF）的空间相关约束NMF线性解混算法（MRF-NMF）。首先,通过基于最小误差的高光谱信号识别（HySime）法估算端元数量,同时利用顶点成分分析（VCA）和全约束最小二乘法（FCLS）初始化端元矩阵与丰度矩阵;其次,利用MRF模型建立描述地物空间分布规律的能量函数,以此描述地物分布的空间相关特征;最后,将基于MRF的空间相关约束函数与NMF标准目标函数以交替迭代的形式参与解混,得出高光谱数据的端元信息与丰度分解结果。理论分析和真实数据实验结果表明,在高光谱数据空间相关程度较低的情况下,相比最小体积约束的NMF （MVC-NMF）、分段平滑和稀疏约束的NMF （PSNMFSC）和交互投影子梯度非负矩阵分解（APS-NMF）三种参考算法,所提算法的端元分解精度仍分别提高了7.82%、12.4%和10.1%,其丰度分解精度仍分别提高了8.34%、12.6%和9.87%。MRF-NMF能够弥补NMF对于空间相关特征描述能力的不足,减小解混结果中地物的空间能量分布误差。     

基于知识决策树的城市水体提取方法研究

针对城市水体与建筑物阴影、沥青路面和浓密植被等暗地物的光谱混淆性,构建了结合光谱特征和空间特征的城市水体提取知识决策树。其基本思路为：首先 利用短波红外波段提取暗地物,其次分别利用浓密植被在近红外波段和沥青路面在红波段中的反射率剔除这两类暗地物,再次利用空间密度特征剔除建筑物阴影,最 后根据面积对水体进行补充识别。与现有方法相比,本方法提出了城市水体提取中需关注的暗地物类型并开展针对性特征分析,并利用由噪声环境下密度聚类方法 （DBSCAN）描述的空间密度特征区分城市水体和建筑物阴影。对北京城区SPOT 5多光谱影像开展的实验得到的检测率为86.18%,虚警率为13.82%,表明本方法是基于 中分辨率多光谱影像提取城市水体的有效方法。     

Sampling design and uncertainty based on spatial variability of spectral variables for mapping vegetation cover

Optimal sampling design for collecting ground data is critical in order to accurately map vegetation cover using remotely sensed data. Traditional simple random sampling often leads to a duplication of information and to a larger sample than is required. An optimal sampling grid spacing based on regionalized variable theory can greatly reduce the number of sample plots needed given a precision level for a study area. However, this method requires a set of ground data that exists or can be obtained via a pilot survey in order to derive a semivariogram for measuring the spatial variability of the variable of interest. In this study, we first developed a method to estimate the semivariogram of a ground or primary variable—vegetation cover from remotely sensed data instead of ground data—and then used it for determining optimal grid spacing for sampling the primary variable. The method developed can avoid the need for a pilot survey to obtain a ground dataset that has a good spatial distribution of plots and can be used to calculate the unbiased semivariogram of the ground variable when unbiased historical data are not available. This can reduce the total cost of collection of ground data. The accuracy of mapping vegetation cover based on this approach was compared to that generated with simple random sampling. A simple sensitivity analysis was conducted. The results show that this new method is very promising for determining optimal sampling grid spacing for estimating regional averages. When it is applied to determining sampling grid spacing for local estimation, a high correlation between vegetation cover and spectral variables is required.     

Temperature and emissivity separation from ASTER data for low spectral contrast surfaces

The performance of Advanced Spaceborne Thermal Emission Reflection Radiometer (ASTER) thermal infrared (TIR) data product algorithms was evaluated for low spectral contrast surfaces (such as vegetation and water) in a test site close to Valencia, Spain. Concurrent ground measurements of surface temperature, emissivity, and atmospheric radiosonde profiles were collected at the test site, which is a thermally homogeneous area of rice crops with nearly full vegetation cover in summer. Using the ground data and the local radiosonde profiles, at-sensor radiances were simulated for the ASTER TIR channels and compared with L1B data (calibrated at-sensor radiances) showing discrepancies up to 3% in radiance for channel 10 at 8.3 μm (equivalently, 2.5 °C in temperature or 7% in emissivity), whereas channel 13 (10.7 μm) yielded a closer agreement (maximum difference of 0.5% in radiance or 0.4 °C in temperature). We also tested the ASTER standard products of land surface temperature (LST) and spectral emissivity generated with the Temperature-Emissivity Separation (TES) algorithm with standard atmospheric correction from both global data assimilation system profiles and climatology profiles. These products showed anomalous emissivity spectra with lower emissivity values and larger spectral contrast (or maximum-minimum emissivity difference, MMD) than expected, and as a result, overestimated LSTs. In this work, a scene-based procedure is proposed to obtain more accurate MMD estimates for low spectral contrast materials (vegetation and water) and therefore a better retrieval of LST and emissivity with the TES algorithm. The method uses various gray-bodies or near gray-bodies with known emissivities and assumes that the calibration and atmospheric correction performed with local radiosonde data are accurate for channel 13. Taking the channel 13 temperature (atmospherically and emissivity corrected) as the true LST, the radiances for the other channels were simulated and used to derive linear relationships between ASTER digital numbers and at-ground radiances for each channel. The TES algorithm was applied to the adjusted radiances and the resulting products showed a closer agreement with the ground measurements (differences lower than 1% in channel 13 emissivities and within ± 0.3 °C in temperature for rice and sea pixels).     

Phase correlation pixel‐to‐pixel image co‐registration based on optical flow and median shift propagation

With singular value decomposition (SVD) and robust 2‐dimensional fitting phase correlation algorithms, it is possible to achieve pixel‐to‐pixel image co‐registration at sub‐pixel accuracy via local feature matching. However, the method often fails in featureless and low correlation areas making it not robust for co‐registration of images with considerable spectral differences and large featureless ground objects. A median shift propagation (MSP) technique is proposed to eliminate the problem, in a phase correlation and Normalized Cross‐Correlation (NCC) combined approach. The experiment results using images from different sensor platforms and spectral bands indicate that the new method is very robust to featureless and low correlation areas and can achieve very accurate pixel‐to‐pixel image co‐registration with good tolerance of spectral and spatial differences between images. The method will significantly improve change detection in various remote sensing applications.     

Object-based land-cover change detection applied to Brazilian seasonal savannahs using geostatistical features

A new method for remote-sensing land-use/land-cover (LULC) change detection is proposed to eliminate the effects of forest phenology on classification results. This method is insensitive to spectral changes caused by vegetation seasonality and uses an object-based approach to extract geostatistical features from bitemporal Landsat TM (Thematic Mapper) images. We first create image objects by multiresolution segmentation to extract geostatistical features (semivariogram parameters and indices) and spectral information (average values) from NDVI (normalized difference vegetation index), acquired in the wet and dry seasons, as input data to train a Support Vector Machine algorithm. We also used the image difference traditional change-detection method to validate the effectiveness of the proposed method. We used two classes: (1) LULC change class and (2) seasonal change class. Using the most geostatistical features, the change detection results are considerably improved compared with the spectral features and image differencing technique. The highest accuracy was achieved by the sill (σ² overall variability) semivariogram parameter (95%) and the AFM (area first lag–first maximum) semivariogram index (88.33%), which were not affected by vegetation seasonality. The results indicate that the geostatistical context makes possible the use of bitemporal NDVI images to address the challenge of accurately detecting LULC changes in Brazilian seasonal savannahs, disregarding changes caused by phenological differences, without using a dense time series of remote-sensing images. The challenge of extracting accurate semivariogram curves from objects of long and narrow shapes requires further study, along with the relationship between the scale of segmentation and image spatial resolution, including the type of change and the initial land-cover class.