An approach to multi-temporal MODIS image analysis using image classification and segmentation

This paper discusses an approach for river mapping and flood evaluation based on multi-temporal time series analysis of satellite images utilizing pixel spectral information for image classification and region-based segmentation for extracting water-covered regions. Analysis of MODIS satellite images is applied in three stages: before flood, during flood and after flood. Water regions are extracted from the MODIS images using image classification (based on spectral information) and image segmentation (based on spatial information). Multi-temporal MODIS images from “normal” (non-flood) and flood time-periods are processed in two steps. In the first step, image classifiers such as Support Vector Machines (SVM) and Artificial Neural Networks (ANN) separate the image pixels into water and non-water groups based on their spectral features. The classified image is then segmented using spatial features of the water pixels to remove the misclassified water. From the results obtained, we evaluate the performance of the method and conclude that the use of image classification (SVM and ANN) and region-based image segmentation is an accurate and reliable approach for the extraction of water-covered regions.     

An optimal selection of probability distribution functions for unsupervised land cover classification of PALSAR-2 data

Different types of classification techniques are available in the literature for the classification of Synthetic Aperture Radar (SAR) data into various land cover classes. Various SAR images are available for land cover classification such as ALOS PALSAR (PALSAR-1, PALSAR-2), RADARSAT and ENVISAT. In this paper, we have attempted to explore probability distribution function (pdf) based land cover classification using PALSAR-2 data. Over 20 different statistical distribution functions are analyzed for different classes based on statistical parameters. Probability distribution functions are selected based on Chi-squared goodness of fit test for each individual class. A decision tree based classifier is developed for classification based on the selected pdf functions and its statistical parameters. The proposed classification approach has an accuracy of 83.93%.     

Assessment of the impact of dimensionality reduction methods on information classes and classifiers for hyperspectral image classification by multiple classifier system

Identification of the appropriate combination of classifier and dimensionality reduction method has been a recurring task for various hyperspectral image classification scenarios. Image classification by multiple classifier system has been evolving as a promising method for enhancing accuracy and reliability of image classification. Because of the diversity in generalization capabilities of various dimensionality reduction methods, the classifier optimal to the problem and hence the accuracy of image classification varies considerably. The impact of including multiple dimensionality reduction methods in the MCS architecture for the supervised classification of a hyperspectral image for land cover classification has been assessed in this study. Multi-source airborne hyperspectral images acquired over five different sites covering a range of land cover categories have been classified by a multiple classifier system and compared against the classification results obtained from support vector machines (SVM). The MCS offers acceptable classification results across the images or sites when there are multiple dimensionality reduction methods in addition to different classifiers. Apart from offering acceptable classification results, the MCS indicates about 5% increase in the overall accuracy when compared to the SVM classifier across the hyperspectral images and sites. Results indicate the presence of dimensionality reduction method specific empirical preferences by land cover categories for certain classifiers thereby demanding the design of MCS to support adaptive selection of classifiers and dimensionality reduction methods for hyperspectral image classification.     

Crater detection via genetic search methods to reduce image features

Recent approaches to crater detection have been inspired by face detection’s use of gray-scale texture features. Using gray-scale texture features for supervised machine learning crater detection algorithms provides better classification of craters in planetary images than previous methods. When using Haar features it is typical to generate thousands of numerical values from each candidate crater image. This magnitude of image features to extract and consider can spell disaster when the application is an entire planetary surface. One solution is to reduce the number of features extracted and considered in order to increase accuracy as well as speed. Feature subset selection provides the operational classifiers with a concise and denoised set of features by reducing irrelevant and redundant features. Feature subset selection is known to be NP-hard. To provide an efficient suboptimal solution, four genetic algorithms are proposed to use greedy selection, weighted random selection, and simulated annealing to distinguish discriminate features from indiscriminate features. Inspired by analysis regarding the relationship between subset size and accuracy, a squeezing algorithm is presented to shrink the genetic algorithm’s chromosome cardinality during the genetic iterations. A significant increase in the classification performance of a Bayesian classifier in crater detection using image texture features is observed.     

基于小波包分解和FCM聚类的纹理图像分割方法

提出了一种新的图像特征提取中选取最优小波分解树的方法.塔式小波分解对信号解不够全面,而小波包全分解又引入庞大的计算量,因此小波分解最优树的选取尤为重要.结合模糊c均值(FCM,Fuzzy C-Mean)聚类,提出了一种能同时进行小波自适应分解和纹理特征分类的纹理图像分割方法,该方法将无监督聚类中的聚类有效性参数引入到自适应小波分解的判决中,能根据无监督聚类分割的需要,自适应地选取小波包分解的树形结构和分解层数.相对于小波包全分解,节省了大量的运算,并能取得良好的分割效果.      

An improved land cover classification using polarization signatures for PALSAR 2 data

Land cover classification in mixed land cover scenarios is challenging with PolSAR data. Polarimetric decomposition techniques are most popular methods for PolSAR data classification in recent times. These techniques focus on identification of dominant scattering phenomena and hence result in sub-optimal classification in mixed land cover scenarios. Alternatively, polarization signatures (PSs) are good illustrations of SAR target responses as they depict a detailed physical information from target backscatter. Researchers have successfully utilized SAR PSs for land cover (LC) classification. Some reports suggested utilizing correlation between observed PSs and standard target PSs as features for LC classification. This paper presents a study on improved utilization of PSs for optimal LC classification in mixed class scenarios. First, PS based SAR features are derived using fully polarimetric SAR data. The features represent a degree of similarity between observed and standard PSs. The derived features are termed as polarization signatures correlation features or PSCFs. The novel PSCFs are analyzed, evaluated and compared with decomposition based features for the purpose of LC classification. Classification performance indicators highlight potential of PSCFs for mixed LC classification problems. Therefore, further an adaptive and optimal LC class boundary estimation approach for LC classification is proposed and developed. Observed PSs and reference LC class PS statistics are used to build empirical models between classification performance indicators and LC class boundaries. The empirical models are optimized using the evolutionary genetic algorithm to maximize classification performance. A decision tree is constructed based on the optimal class boundaries to prepare LC classification. The proposed classification approach is compared with some recent popular classifiers and comparison suggests that the proposed approach provides satisfactory results for mixed LC classification scenarios.     

Land use/land cover classification using time series Landsat 8 images in a heavily urbanized area

It is of great significance to timely, accurately, and effectively monitor land use/cover in city regions for the reasonable development and utilization of urban land resources. The remotely sensed dynamic monitoring of Land use/land cover (LULC) in rapidly developing city regions has increasingly depended on remote-sensing data at high temporal and spatial resolutions. However, due to the influence of revisiting periods and weather, it is difficult to acquire enough time-series images with high quality at both high temporal and spatial resolution from the same sensor. In this paper we used the temporal-spatial fusion model ESTARFM (Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model) to blend Landsat8 and MODIS data and obtain time-series Landsat8 images. Then, land cover information is extracted using an object-based classification method. In this study, the proposed method is validated by a case study of the Changsha City. The results show that the overall accuracy and Kappa coefficient were 94.38% and 0.88, respectively, and the user/producer accuracies of vegetation types were all over 85%. Our approach provides an accurate and efficient technical method for the effective extraction of land use/cover information in the highly heterogeneous regions.     

Selection of classification techniques for land use/land cover change investigation

The concerns over land use/land cover (LULC) change have emerged on the global stage due to the realisation that changes occurring on the land surface also influence climate, ecosystem and its services. As a result, the importance of accurate mapping of LULC and its changes over time is on the increase. Landsat satellite is a major data source for regional to global LULC analysis. The main objective of this study focuses on the comparison of three classification tools for Landsat images, which are maximum likelihood classification (MLC), support vector machine and artificial neural network (ANN), in order to select the best method among them. The classifiers algorithms are well optimized for the gamma, penalty, degree of polynomial in case of SVM, while for ANN minimum output activation threshold and RMSE are taken into account. The overall analysis shows that the ANN is superior to the kernel based SVM (linear, radial based, sigmoid and polynomial) and MLC. The best tool (ANN) is then applied on detecting the LULC change over part of Walnut Creek, Iowa. The change analysis of the multi temporal images indicates an increase in urban areas and a major shift in the agricultural practices.     

A method in near-surface estimation of air temperature (NEAT) in times following the satellite passing time using MODIS images

Air temperature is one of the most important parameters in environmental, agricultural and water resources studies. This information is not usually always available at the required temporal and spatial resolution. The air temperature is measured at a fixed point in the meteorological stations which are dispersed and may not have the appropriate spatial resolution needed for many applications. On the other hand, MODIS satellite images have relatively acceptable spatial resolution specially for use in environmental studies. There is a methodology with which the near surface air temperature can be extracted from MODIS images at the satellite passing time with an acceptable accuracy. The goal in this study is to find a way to predict the air temperature in times after/before the satellite passing time. The procedure consists of two steps. In the first step, the relationship between the air temperature at a time in a synoptic station and the air temperature in other times up to 5 h later were modeled. In the second step, using these built up relationships, the air temperature extracted from the satellite image at the passing time was extrapolated to the next hours. Finally, the results of this extrapolation method were evaluated using the air temperatures measured at those hours and in the pixels containing some other meteorological stations. The error of the method when applied to a relatively homogeneous surface cover was about 1.5 °C. This error when applied to the next hours, was below 2 °C up to 5 h after satellite passing time. This method can be useful in some agricultural and horticultural applications in which both the spatial and temporal resolution are needed simultaneously. This product is a useful tool for frost prediction, a phenomenon that usually happens at night or early in the morning.     

Advanced fractal approach for unsupervised classification of SAR images

Unsupervised classification of Synthetic Aperture Radar (SAR) images is the alternative approach when no or minimum apriori information about the image is available. Therefore, an attempt has been made to develop an unsupervised classification scheme for SAR images based on textural information in present paper. For extraction of textural features two properties are used viz. fractal dimension D and Moran’s I. Using these indices an algorithm is proposed for contextual classification of SAR images. The novelty of the algorithm is that it implements the textural information available in SAR image with the help of two texture measures viz. D and I. For estimation of D, the Two Dimensional Variation Method (2DVM) has been revised and implemented whose performance is compared with another method, i.e., Triangular Prism Surface Area Method (TPSAM). It is also necessary to check the classification accuracy for various window sizes and optimize the window size for best classification. This exercise has been carried out to know the effect of window size on classification accuracy. The algorithm is applied on four SAR images of Hardwar region, India and classification accuracy has been computed. A comparison of the proposed algorithm using both fractal dimension estimation methods with the K-Means algorithm is discussed. The maximum overall classification accuracy with K-Means comes to be 53.26% whereas overall classification accuracy with proposed algorithm is 66.16% for TPSAM and 61.26% for 2DVM.     

Surface temperature estimation in Singhbhum Shear Zone of India using Landsat-7 ETM+ thermal infrared data

Land surface temperature (LST) is an important factor in global change studies, heat balance and as control for climate change. A comparative study of LST over parts of the Singhbhum Shear Zone in India was undertaken using various emissivity and temperature retrieval algorithms applied on visible and near infrared (VNIR), and thermal infrared (TIR) bands of high resolution Landsat-7 ETM+ imagery. LST results obtained from satellite data of October 26, 2001 and November 2, 2001 through various algorithms were validated with ground measurements collected during satellite overpass. In addition, LST products of MODIS and ASTER were compared with Landsat-7 ETM+ and ground truth data to explore the possibility of using multi-sensor approach in LST monitoring. An image-based dark object subtraction (DOS3) algorithm, which is yet to be tested for LST retrieval, was applied on VNIR bands to obtain atmospheric corrected surface reflectance images. Normalized difference vegetation index (NDVI) was estimated from VNIR reflectance image. Various surface emissivity retrieval algorithms based on NDVI and vegetation proportion were applied to ascertain emissivities of the various land cover categories in the study area in the spectral range of 10.4–12.5 μm. A minimum emissivity value of about 0.95 was observed over the reflective rock body with a maximum of about 0.99 over dense forest. A strong correlation was established between Landsat ETM+ reflectance band 3 and emissivity. Single channel based algorithms were adopted for surface radiance and brightness temperature. Finally, emissivity correction was applied on ‘brightness temperature’ to obtain LST. Estimated LST values obtained from various algorithms were compared with field ground measurements for different land cover categories. LST values obtained after using Valor’s emissivity and single channel equations were best correlated with ground truth temperature. Minimum LST is observed over dense forest as about 26 °C and maximum LST is observed over rock body of about 38 °C. The estimated LST showed that rock bodies, bare soils and built-up areas exhibit higher surface temperatures, while water bodies, agricultural croplands and dense vegetations have lower surface temperatures during the daytime. The accuracy of the estimated LST was within ±2 °C. LST comparison of ASTER and MODIS with Landsat has a maximum difference of 2 °C. Strong correlation was found between LST and spectral radiance of band 6 of Landsat-7 ETM+. Result corroborates the fact that surface temperatures over land use/land cover types are greatly influenced by the amount of vegetation present.     

深度学习机制与小波融合的超分辨率重建算法

深度学习技术在超分辨率重建领域中发展迅速。为了进一步提升重建图像的质量和视觉效果，针对基于生成对抗网络（GAN）的超分辨率重建算法重建图像的纹理放大后不自然的问题，提出了一种结合小波变换和生成对抗网络的超分辨率重建算法。所提算法在生成对抗网络中将小波分解的每个分量在各自独立的子网中进行训练，实现网络对小波系数的预测，有效地重建出具有丰富的全局信息和局部纹理细节信息的高分辨率图像。实验结果表明，对比基于生成对抗网络的算法，所提算法重建图像的客观评价指标峰值信噪比（PSNR）和结构相似性分别能提高至少0.99 dB和0.031。      

基于小波变换和神经网络的脑电信号分类方法

结合小波变换和神经网络对酒精中毒者和正常清醒者的脑电信号进行分类.通过分析脑电数据找出分类特征;采用一维离散小波变换提取含有分类特征的脑电信号频段,并以小波变换分解系数作为信号特征,实现数据序列长度压缩;对应3种刺激方式建立3个相同结构的学习向量量化(LVQ)神经网络,用于对脑电信号的预分类;根据判决规则得到最终分类结果.对真实脑电数据的分类正确率达到89%.     

基于卫星遥感的岛礁影像多类目标智能化提取

卫星遥感影像具有背景复杂、目标尺度不一、观测方向各异、纹理不清晰等特点，主流的深度学习目标检测算法不能直接适用于卫星遥感影像的目标检测。改进了RetinaNet，使其适用于卫星遥感影像。首先设计了一种新的特征融合方式，融合ResNet50输出的特征图，使得融合后的特征图同时具有高层语义信息和低层纹理细节信息。为了减弱遥感影像复杂背景对目标特征的影响，设计了特征感知模块，在减弱噪声对特征图影响的同时增强有用特征。挑选DOTA数据集中船只、飞机和存储罐图像进行训练和测试。改进的算法与RetinaNet相比，飞机、船只和存储罐的平均精度分别提高了41%、25%、24%。基于高分二号卫星（GF 2）真实影像数据的试验结果表明，提出的算法能够用于卫星遥感岛礁影像的多类目标智能化提取。     

Simulation of the land surface temperature from moon-based Earth observations

The land surface temperature (LST) is a key parameter for the Earth’s energy balance. As a natural satellite of the Earth, the orbital of the moon differs from that of current Earth observation satellites. It is a new way to measure the land surface temperature from the moon and has many advantages compared with artificial satellites. In this paper, we present a new method for simulating the LST measured by moon-based Earth observations. Firstly, a modified land-surface diurnal temperature cycle (DTC) method is applied to obtain the global LST at the same coordinated universal time (UTC) using the Moderate Resolution Imaging Spectroradiometer (MODIS) LST products. The lunar elevation angles calculated using the ephemeris data (DE405) from the Jet Propulsion Laboratory (JPL) were then applied to simulate the Earth coverage observed from the moon. At the same time, the modified DTC model was validated using in situ data, MODIS LST products, and the FengYun-2F (FY-2F) LST, respectively. The results show that the fitting accuracy (root-mean-square error, RMSE) of the modified DTC model is not greater than 0.72?°C for eight in situ stations with different land cover types, and the maximum fitting RMSE of the modified model is smaller than that of current DTC models. By the comparison of the simulated LST with MODIS and FY-2F LST products, the errors of the results were feasible and accredited, and the simulated global LST has a reasonable spatiotemporal distribution and change trend. The simulated LST data can therefore be used as base datasets to simulate the thermal infrared imagery from moon-based Earth observations in future research.     

基于对抗和迁移学习的灾害天气卫星云图分类

针对卫星云图中的灾害天气数据存在严重不平衡问题，提出一个结合生成对抗学习（GAN）和迁移学习（TL）的卷积神经网络（CNN）框架以解决上述问题进而提高基于卫星云图的灾害天气分类精度。该框架主要包含基于GAN的数据均衡化模块和基于迁移学习的CNN分类模块。上述2个模块分别从数据和算法层面解决数据的类间不平衡问题，分别得到一个相对均衡的数据集和一个可在不同类别数据上提取相对均衡特征的分类模型，最终实现对卫星云图的分类，提高其中灾害天气的卫星云图类别分类准确率。与此同时所提方法在自建的大规模卫星云图数据上进行了测试，消融性和综合实验结果证明了所提数据均衡方法和迁移学习方法是有效的，且所提框架模型对各个灾害天气类别的分类精度都有显著提升。      

基于小波变换熵值及高阶累积量联合的卫星信号调制识别算法

调制识别是信号检测与解调的关键环节，针对卫星调制中采用的MAPSK，MQAM，MFSK，MPSK方式，提出了一种计算小波变换熵值并结合高阶累积量的联合调制识别算法.根据小波变换对时频信息敏感的特点，不同调制方式高阶累积量计算结果的区分性以及不同复杂度的调制信号熵值结果不同，分析了以上4类调制信号的计算结果，提出了基于小波变换熵值及高阶累积量联合的卫星信号调制识别算法.计算调制信号小波系数，据此计算熵值，实现对调制信号的类别划分，使用高阶累积量实现类别内的信号分类.经过仿真分析，可实现在8dB以上达到0.9识别率的效果，该方法对高阶（64阶调制）信号识别具有一定借鉴意义.      

Spatial and temporal scales of the Brazil–Malvinas Current confluence documented by simultaneous MODIS Aqua 1.1-km resolution SST and color images

The moderate resolution imaging spectroradiometer (MODIS) on board the Aqua satellite measures visible and infrared radiation in 36 wavebands, providing simultaneous images of sea-surface temperature (SST) and chlorophyll-a (chl-a) concentration in the upper meters of the sea. For the first time, truly synoptic SST and chl-a- concentration images are available. These images are daily and of 1.1-km resolution.The strong contrasts in sea-surface temperature and surface chlorophyll-a concentration over the southwest Atlantic make satellite infrared and color images particularly appropriate tools for studying the Brazil–Malvinas (B/M) Current confluence. We examine two years (July, 2002–June, 2004) of Aqua/MODIS infrared and color images to document the precise structure of the B/M confluence simultaneously in SST and chl-a.We first compared MODIS weekly data with simultaneous independent satellite data. Spatial and temporal distributions are similar for both SST and color. Differences between MODIS and SeaWiFS (sea-viewing wide field-of-view sensor) are large in pigment-rich regions along the coast and shelf. Here, we focused on the offshore region where differences are small.For each season, exceptionally cloud-free 1.1-km resolution MODIS images showed two thermal fronts, one corresponding to the Brazil Current’s southernmost limit, the other, to the Malvinas Current’s northernmost limit. These two fronts remained quite close to each other (within 50 km) and were separated by water with an SST and chl-a concentration typical of the continental shelf waters. In spring, the water rich in chl-a from the platform is squeezed between the two currents and entrained away from the coast in between the two thermal fronts. In the frontal region, SST gradient maxima trace the contour of the chl-a-rich water.Enlargements of the frontal region and of the turbulent region downstream of the frontal collision are presented and analyzed. MODIS documents in an unprecedented way the SST and chl-a filaments as they are distorted and mixed by meso- and sub-mesoscale structures in the strain-dominated region of the B/M confluence. It is suggested that a substantial part of the chl-a local maximum in the Malvinas return flow is of continental-shelf origin.     

基于免疫抗体编码机理的卫星图像云探测

针对卫星遥感图像中云与下垫面的复杂性和多样性,提出了一种新的在复杂背景下常用目标图像描述方法.通过类比生物免疫抗体特异性与其构成单元氨基酸性质的相关性,得到目标图像的免疫基元集合形式及其分类方法.借鉴生物免疫抗体编码顺序中氨基酸结合能量最小原则,统计分析出训练样本的免疫基元亲和度计算公式,实现了目标图像描述的有限多特征优化组合.对于云检测问题,最终提出云的免疫抗体编码方法,定义了云的免疫亲和度计算公式,并成功构建了云免疫抗体.经200幅IKONOS卫星图像测试,验证了该方法在识别率和运行时间效率方面的有效性.