In the field of Remote Sensing Scene Classification (RSSC), multi-label classification has become necessary. However, the creation of a multi-label dataset is a laborious process due to the higher annotation costs compared to multi-class classification. In this study, we conducted a pioneering experiment in the context of partial-label classification on remote sensing datasets and aim to discuss the differences and limitations. In partial-label classification, each image is assigned some "positive" labels, which means it is annotated, and other "unknown" labels which are not determined as positive or negative. Consequently, the model is trained with limited information. We evaluated the classification performance on MLRSNet and AID multilabel datasets, using the method and loss functions that have shown excellent performance in previous studies on ground-level view datasets. Our code is available at https://github.com/Kf-7070/ IGARSS2023_partial_label.

In the field of remote sensing, there has been a significant amount of research focused on linking different domains such as multi-resolution, multi-spectral, and multi-sensor imagery. One task that involves such multimodal data is cross-view geo-localization, which aims to identify the location of ground query images by matching them with aerial images in a database that is tagged with GPS information. Our study presents a novel lightweight convolutional neural network architecture that achieves comparable performance to a transformer-based model on a city-scale dataset while reducing the number of parameters without using any data augmentation and transformation. Furthermore, our experimental findings indicate that the presence or absence of the fully-connected layer, which is used for generating attention maps, has minimal influence on the model’s accuracy.The code is available from https://github.com/ryotayagiABC/Light_CVGL.

Convolutional Neural Networks (CNN) were recently applied to stereo matching, which is one of the important steps in 3D reconstruction. In remote sensing using Very High Resolution (VHR) satellite imagery, CNN-based stereo matching is employed in Digital Surface Models (DSM) generation, achieving higher accuracy than conventional methods through supervised learning. Since ground-truth disparity maps calculated from reliable elevation data are hard to collect, unsupervised stereo matching methods are as significant as supervised ones. However, there are few studies on stereo matching of VHR satellite imagery with unsupervised learning. In this work, we propose an unsupervised stereo matching network for DSM generation. We also introduce a criterion to select the best epoch without using ground-truth data for validation in a training strategy of gradually increasing the weight of a smoothness loss. Experimental results show that our network performs better in the average endpoint error and the fraction of erroneous pixels than the baseline method of the used dataset without ground-truth data.

Hyperspectral Imager Suite (HISUI) was launched in December 2019 and started its operation observation from International Space Station in October 2020. HISUI data are being acquired based on data acquisition requests from HISUI project and principal investigators of HISUI Research Announcement. The total area of acquired HISUI data by December 2021 exceeded 110 M km2and the data are being used in various application studies. HISUI will be operated by FY2023. In this presentation, HISUI's instrument and operation status will be presented together with its early results in terms of geology and greenhouse gas emission.

This paper reports on the initial analysis of spectral smile calibration of the Hyperspectral Imager Suite (HISUI) onboard the International Space Station, which has been continuously acquiring data since September 4, 2020. HISUI is an optical hyperspectral imager consisting of two subsystems: VNIR covering 400 to 980nm at intervals of 10 nm, and SWIR covering 895 to 2481nm at 12.5nm intervals. Based on the atmospheric correction for actual observation images, we assessed cross-track dependences of the wavelength deviation (spectral smile) and the full-width at half-maximum (FWHM) of the HISUI response function. We found that significant spectral smile was observed, with maximum variations of 1.8nm in VNIR and 4.3–4.5nm in SWIR. In addition, the cross-track variation of FWHM was observed with maximum variations of 5.0nm for VNIR and 2.5–3.5nm for SWIR.We used the results to model the smile functions to update a smile correction table in the internal calibration system of HISUI. Then, we evaluated how the smile functions reduce the spectral smile in the data acquired after the update on September 27, 2021.We confirmed that VNIR showed a nearly flat profile within 0.25nm with a nearly constant FWHM. For SWIR, although a slight amount of spectral smile and a variation of FWHM were still observed partly due to wavelength dependence in the spectral smile, the spectral smile was reduced to <~2.2 nm. This study demonstrated that wavelength calibration using actual observation images for ground surfaces is important for the characterization of hyperspectral sensors.

Hyperspectral Imager Suite (HISUI) was launched in December 2019 and successfully attached to Japan Experiment Module Exposed Facility, International Space Station. Due to data communication problems, its initial check out activities have been delayed, but its first light image was acquired and released in September 2020. HISUI data delivery to Research Users will start in 1Q of Japanese FY2021 (April - June 2021).

Since synthetic-aperture radar (SAR) imagery is not susceptible to weather and sunlight conditions, it is suitable for disaster mapping. However, ionospheric refraction of microwaves influence the registration of SAR imagery and reduces mapping accuracy, consequently hindering quick rescue operations. In this work, a registration method based on phase correlation was applied for SAR intensity images to map landslides caused by heavy rain in Japan. The mapping accuracy improved by decreasing false positives via co-registration. A specific pattern of intensity changes was observed around collapses, which is attributable to topographic changes due to landslides. The pattern reduced false positives further when used in pattern matching.

HISUI, the Japanese hyperspectral sensor, was launched on December 6, 2019 and the first image was taken on September 4, 2020. The first HISUI calibration was conducted on September 11. Wavelength and radiance calibration were conducted to evaluate characteristic change during the launch process. We found 2.3 nm and 5.5 nm wavelength shifts for VNIR and SWIR, respectively using SRM2025a filter onboard HISUI calibration system. Sensitivity changes during the launch process were less than 1.0 % that is smaller than the radiometric accuracy specifications.

To realize the pixel-based classification of hyperspectral data, wavelet transform is applied to produce two-dimensional data, which makes it possible to combine convolutional neural networks (CNNs). Although wavelet transform is usually used for time series data, it can also be applied to hyperspectral data which consist of many bands. By analyzing hyperspectral data with multiple resolutions, the characteristics of the spectrum can be captured on various scales. The features generated by the wavelet transform are used as input data to the CNN classifier to identify class-specific patterns. We verified this approach using two data sets and confirmed its effectiveness. Futhermore, the proposed method performed well even with a small number of training samples.

High stability of the observation satellite platform is increasingly demanded in recent years. However, attitude jitter of observation satellites is a problem that degenerates the development of imaging quality and resolution. In order to reduce the geo-positioning errors and improve the geometric accuracy of remote sensing images, satellite jitter have been studied in recent years. In this work, a generative adversarial network (GAN) architecture is proposed to automatically learn and correct the deformed scene features from a single remote sensing image. In the proposed GAN, a convolutional neural network (CNN) is designed to discriminate the inputs and another CNN is used to generate so-called fake inputs. In order to explore the usefulness and effectiveness of GAN for jitter detection, the proposed GAN are trained on part of PatternNet dataset and tested on three popular remote sensing datasets. Several experiments show that the proposed models provide competitive results compared to other methods. the proposed GAN reveals the huge potential of GAN-based methods for the analysis of attitude jitter from remote sensing images.

Deep learning-based approaches for land cover segmentation rely on supervision with pixel-level ground truth, but may not generalize well to unseen image domains. Due to the tedious and labor-intensive labeling process, transferring the models trained with label-rich source data to nonannotated target data becomes a popular problem in recent years. Because of the domain shift, the difference between the source and target distributions can degrade the accuracy on target data if the training occurs directly in a source domain without proper domain adaptation (DA). In this letter, we propose a U-Net based network for DA in the context of semantic segmentation. The model is trained in the source domain with ground truth and test in the target domain without any annotations. We introduce the layer alignment method and the feature covariance loss function to alleviate the domain shift between different domains. To further enhance the adapted model, we adopt a self-training method to improve segmentation performance. Experimental results on the images from the 2018 IEEE Geoscience and Remote Sensing Society (GRSS) data fusion contest and the International Society for Photogrammetry and Remote Sensing (ISPRS) 2-D semantic labeling contest data set reveal the effectiveness of the proposed model. By reducing the domain distribution difference, our method shows better performance compared with the mainstream unsupervised DA methods.

In this paper, we propose a SAR-to-optical image translation method based on conditional generative adversarial networks (cGANs). Though cGANs have achieved great success in image translation, some problems remain in SAR-to-optical image translation. One of the problems is the colorization error owing to the lack of color information in SAR data. Since the colors of optical images are varied, while SAR images have no color information, the generator network is confused and fail to generate correctly colorized optical images. To prevent it, we introduce a region information to the image translation network. Specifically, the feature vector from the pre-trained classification network is fed to the generator and discriminator network. Experimental results with SEN1-2 dataset show the advantage of our proposed method over the baseline method that does not use any additional information.

Farmland in the Fukushima region of Japan experienced serious radioactive contamination as a result of the Fukushima Nuclear Power Plant disaster in 2011. A large number of farm-fields have been decontaminated by replacing the top surface soil with non-contaminated soil. This preliminary study investigated the potential of high spatial- and temporal-resolution micro-satellite Venμs and ground-based hyperspectral sensing techniques for assessment of soil fertility and rice productivity in the region to assist the agricultural recovery. Results suggested that these variables would be estimated in a regional scale by remote sensing approaches.

This work presents an implementation of Deepmatching to detect the displacement of faults caused by large earthquakes. Deepmatching uses the global information of the input images as well as the local information near the patch to find pairs of image patches. This work solves two problems of Deepmatching in terms of remote sensing applications. First, dense matching results of Deepmatching are obtained to directly detect faults caused by earthquakes. Second, the reduction in computational cost of Deepmatching is achieved so that satellite imagery, which is generally larger, is processed. As a result, our extended Deepmatching successfully calculates the surface displacement map. Finally, we applied the proposed method to the Balochistan earthquake and successfully detected the faults caused by it.

Hyperspectral Imager Suite (HISUI) was launched on Dec. 5, 2019 and successfully attached to Japan Experiment Module Exposed Facility, International Space Station. Currently, initial check out activities are ongoing, and the first light images will be released around August, 2020. The distribution of HISUI Level 1 products will be started in the first half of 2021.

Since hyperspectral data is composed of many spectral bands, it is used for classification with high accuracy. Various studies using spatial information as well as spectral information have been conducted to improve accuracy. However, in recent years pixel-based methods have been reviewed for purposes such as anomaly detection, which is originally expected for hyperspectral data. In this work, we applied one of the time-frequency analysis techniques, wavelet transform, to improve the information extraction capability from hyperspectral data in both classification and regression tasks. As the result, the proposed method showed higher accuracy for classification compared to the conventional methods without spatial information. Also, it was confirmed that the proposed method was effective even for small size classes. For regression, we compared various regression models and confirmed the effectiveness of the proposed method for almost all models.

The advent of satellite constellation has been rapidly increasing the acquisition frequency of remote sensing imagery. In spite of widespread use of constellation satellite imagery in various applications, the assessments of these imagery are still limited. To assess the quality of constellation imagery, we compared imagery of the constellation satellite, Planet, with imagery of the conventional satellite, VENμS, by using three methodologies. Our research revealed that constellation satellites have the advantage of being less blured and the disadvantage of being noisy. Also, we denoted that relative indices, such as NDVIs, should be used in data fusion application. Based on these analysis, we performed land cover mapping of fused time-series satellite imagery by making a use of the result of these assessments.

Hyperspectral image unmixing has proven to be a useful technique to interpret hyperspectral data, and is a prolific research topic in the community. Most of the approaches used to perform linear unmixing are based on convex geometry concepts, because of the strong geometrical structure of the linear mixing model. However, many algorithms based on convex geometry are still used in spite of the underlying model not considering the intra-class variability of the materials. A natural question is to wonder to what extent these concepts and tools (Intrinsic Dimensionality estimation, endmember extraction algorithms, pixel purity) are still relevant when spectral variability comes into play. In this paper, we first analyze their robustness in a case where the linear mixing model holds in each pixel, but the endmembers vary in each pixel according to a prescribed variability model. In the light of this analysis, we propose an integrated unmixing chain which tries to adress the shortcomings of the classical tools used in the linear case, based on our previously proposed extended linear mixing model. We show the interest of the proposed approach on simulated and real datasets.

Recently, convolutional neural network (CNN) has achieved great results in pansharpening. Most pansharpening methods with CNN are based on PNN [1] inspired by super-resolution methods with CNN and learn the pansharpening of downsampled images. In this work, we presented a novel framework for pansharpening based on two desired property of pansharpened images: downsampled pansharpened images become low-resolution multi-spectral images (spectral consistency) and panchromatic images are approximated by weighted addition of each bands of pansharpened images (spatial consistency). Our framework train CNN to learn this spectral-spatial consistency. The advantage of our framework is that there is no scale mismatch between training and test data. We applied our method to Landsat-8 images and compared it with some previous methods.

Hyperspectral Imager Suite (HISUI) is a Japanese hyperspectral mission to obtain optical images ranging from visible to shortwave infrared region. The mission instrument will be on-board the Japanese Experiment Module (JEM) of the International Space Station (ISS) in early 2020. The sensor obtains spectral images of 185 bands with the ground sampling distance of 20x31 meter, which succeeds the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) in the visible and shortwave infrared region. After the critical design review in 2014, integration of the Flight Model (FM) of HISUI sensor was completed in the beginning of 2017. In parallel, the bus system connecting HISUI sensor and the JEM External Facility (EF) is developed, which has the function of a data recorder, electric control, navigation, thermal control and mechanical structure is under development. Since hyperspectral data needs an accurate calibration, the ground-based test is under way. In this work, the present status of HISUI sensor is presented. Sensor performances in terms of optical distortion and radiometric performances are described based on the FM tests.

To mitigate human damage by a catastrophic disaster, such as Great East Japan Earthquake and Tsunami happened in 2011, a quick satellite observation and data distribution are the keys to recognize damaged area and make plans to rescue. Geosynchronous orbit is the best solution to provide an instantaneous observation from the occurrence of the disaster. However, we could not materialize such satellite system for its large and heavy telescope considering its reasonable ground resolution requirement, such as "less than 10m". Recent progress of and segmented mirror telescope for astronomy on the ground, might make it possible to have such design solution. A system concept study was carried out by JAXA. We designed 3.6m diameter segmented telescope using active/adaptive optics to control wavefront error. The target ground resolution might be 6-8m with visible panchromatic band at Nadir. Also, the target observation latency is less than 30 minutes from receiving requirement to saving geometrically corrected data on server. The large telescope system will be mounted on current satellite bus or all electric propulsion bus and launched by new H3 Launch Vehicle. We are planning for three years' test campaign using partial model of the active mirror system.

Hyperspectral Imager Suite (HISUI) is a future spaceborne hyperspectral Earth imaging system being developed by Japanese Ministry of Economy, Trade, and Industry (METI). HISUI will be launched and deployed on International Space Station (ISS) for two - three year operation from 2020. Manufacturing and testing of Hyperspectral sensor were completed. Integration and test of HISUI Exposed Payload System is being conducted. HISUI data policy and research announcement as well as synergy with other earth observing instruments onboard ISS are also being discussed.

The advent of satellite constellations has been rapidly increasing the acquisition frequency of Earth observation images, enabling continuous updates of land cover maps on a global scale with supervised learning. Since annotation of land cover semantic classes is expensive with a field survey, the number of labeled samples and the update frequency are limited. To compensate for the lack of ground-truth labels, we propose to substitute automatically labeled ground-shot images for human annotated data. In our methodology, unlabeled georeferenced ground-shot images are classified with a Convolutional Neural Network (CNN), which is trained by a set of ground-shot images annotated by humans in advance. We propose a new method that automatically grades human annotation to address noisy labels used for training the CNN. The outcomes of ground-shot image classification are used as labeled samples for pixel-wise land cover classification of multispectral satellite imagery. The resulted land cover classification map resembles the results supervised by human-annotated samples. The experimental results show that our methodology has the potential to associate satellite imagery with ground-shot imagery.

Attitude jitter of observation satellites is a problem that degenerates the imaging quality in high-resolution (HR) remote sensing. In order to retrieve jitters with high accuracy, a new way to fuse image registration and convolutional neural network by a Kalman filter is described. In this method, a novel and complete framework based on image registration are proposed to estimate the on-orbit attitude variations from multi-spectrum remote sensing images. Then a convolutional neural network(CNN) is proposed to estimate the attitude jitter by automatically learning essential scene features from a single image. A Kalman filter is derived to fuse the jitter obtained from the two algorithms. The proposed methodology is trained and tested with PatternNet dataset. Simulation results show that the attitude variations are detected with high accuracy and the deformed images are compensated.

Attitude jitter of satellites and unmanned aerial vehicle (UAV) platforms is a problem that degenerates the imaging quality in high-resolution remote sensing. This letter proposes a deep learning architecture that automatically learns essential scene features from a single image to estimate the attitude jitter, which is used to compensate deformed images. The proposed methodology consists of a convolutional neural network and a jitter compensation model. The neural network analyzes the deformed images and generates the attitude jitter vectors in two directions, which are utilized to correct the images through interpolation and resampling. The PatternNet and the small UAV data sets are introduced to train the neural network and to validate its effectiveness and accuracy. The compensation results on distorted remote sensing images obtained by satellites and UAVs reveal that the image distortion due to attitude jitter is clearly reduced and that the geometric quality is effectively improved. Compared to the existing methods that primarily rely on sensor data or parallax observation, no auxiliary information is required in our framework.

Hyperspectral image unmixing is a source separation problem aiming at recovering the spectra of the pure materials of the observed scene (called endmembers), as well as their relative proportions in each pixel of the image (called abundances). The variability of the materials has recently received a lot of attention in the community. In particular, a consequent number of models and algorithms have been proposed to estimate pixel-wise endmembers to account for their variability. These algorithms often rely on classical endmem-ber extraction algorithms to provide reference spectra. In difficult scenarios with shadows and significant variability these algorithms may fail. In this paper, we address this issue in the Extended Linear Mixing Model framework by considering that an endmember is a direction in the feature space, rather than a single point. Under this paradigm, we show that using k- means clustering with the cosine similarity outperforms geometric endmember extraction algorithms. We also design an algorithm to refine the estimation of the endmember directions, and to account for both illumination and intrinsic variability effects. We show the potential of the proposed algorithm on a synthetic dataset using real world spectra with variability, and a challenging real dataset of a natural scene.

The semantic segmentation of High Resolution Remote Sensing (HRRS) images is the fundamental research area of the earth observation. Convolutional Neural Network (CNN), which has achieved superior performance in computer vision task, is also useful for semantic segmentation of HRRS images. In this work, focal loss is used instead of cross-entropy loss in training of CNN to handle the imbalance in training data. To evaluate the effect of focal loss, we train SegNet and FCN with focal loss and confirm improvement in accuracy in ISPRS 2D Semantic Labeling Contest dataset, especially when γ is 0.5 in SegNet.

Hyperspectral Imager Suite (HISUI) is a future spaceborne hyperspectral Earth imaging system being developed by Japanese Ministry of Economy, Trade, and Industry (METI). HISUI will be launched and deployed on International Space Station (ISS) for two - three year operation from 2020. Manufacturing and testing of HISUI Flight Model were almost completed, and design of HISUI Exposed Payload System is being conducted. HISUI data policy and research announcement as well as synergy with other earth observing instruments onboard ISS are also being discussed.

In this paper, we propose to fuse multiple sources remotely sensed datasets, such as hyperspectral (HS) and Light Detection and Ranging (LiDAR)-derived digital surface model (DSM) using a novel deep learning method. Morphological openings and closings with partial reconstruction are taken into account to model spatial and elevation information for both sources. Then, the stacked features directly input to a deep learning classifier, namely Deep Forest (DF). In particular, Deep Forest can be viewed as the cascade or the ensembles of Rotation Forests (RoF) and Random Forests (RF). We applied the proposed method to the datasets obtained from Tama forest, Japan. Experimental results demonstrate that Deep Forest can achieve better classification results than other approaches. Compared to deep neural networks, deep forest pays little effort in parameter tuning and has a significant reduction in computational complexity.

Domain adaptation and transfer learning adapt the priori information of source domain to train a classier used to predict the label in the target domain. The parameter and instance transfer methods have shown excellent performance. The former adjusts the parameters of transitional classifiers and the latter re-weights the training sample to the different training set, which is similar to the AdaBoost. To further improve the performance, we proposed to combine the two techniques mentioned above. More specifically, we select the Transfer Boosting and domain adaptation extreme learning machine (DAELM) as the instance and parameter transfer methods, respectively. We refer the proposed method to the boosting for DAELM (BDAELM). We compare the proposed method with DAELM and other methods on the real cross-domain hyperspectral remote sensing images acquired over a Japanese mixed forest, showing improved classification accuracies.

Noises in hyper spectral data make it difficult to accurately classify the domains. In order to solve this problem, some filtering methods are proposed; however, filtering processes disturb small size classes and sharp boundaries. To protect the domain boundaries in hyperspectral data classification, we applied normalized cuts segmentation in the preprocessing before classification. The proposed methodologies show higher OA and AA, which means that the domain boundaries are maintained, and classification accuracies of small classes are improved. We also found that the classification accuracies are not sensitive to the number of clusters. The proposed methodology is useful for the combination of smoothing filters that show high classification performances.

Due to the development of sensors and data acquisition technology, the fusion of features from multiple sensors is a very hot topic. In this letter, the use of morphological features to fuse a hyperspectral (HS) image and a light detection and ranging (LiDAR)-derived digital surface model (DSM) is exploited via an ensemble classifier. In each iteration, we first apply morphological openings and closings with a partial reconstruction on the first few principal components (PCs) of the HS and LiDAR data sets to produce morphological features to model spatial and elevation information for HS and LiDAR data sets. Second, three groups of features (i.e., spectral and morphological features of HS and LiDAR data) are split into several disjoint subsets. Third, data transformation is applied to each subset and the features extracted in each subset are stacked as the input of a random forest classifier. Three data transformation methods, including PC analysis, linearity preserving projection, and unsupervised graph fusion, are introduced into the ensemble classification process. Finally, we integrate the classification results achieved at each step by a majority vote. Experimental results on coregistered HS and LiDAR-derived DSM demonstrate the effectiveness and potentialities of the proposed ensemble classifier.

Classification techniques for hyperspectral images based on random forest (RF) ensembles and extended multiextinction profiles (EMEPs) are proposed as a means of improving performance. To this end, five strategies - bagging, boosting, random subspace, rotation-based, and boosted rotation-based - are used to construct the RF ensembles. EPs, which are based on an extrema-oriented connected filtering technique, are applied to the images associated with the first informative components extracted by independent component analysis, leading to a set of EMEPs. The effectiveness of the proposed method is investigated on two benchmark hyperspectral images: the University of Pavia and Indian Pines. Comparative experimental evaluations reveal the superior performance of the proposed methods, especially those employing rotation-based and boosted rotation-based approaches. An additional advantage is that the CPU processing time is acceptable.

Due to the benefits and limitation of different remote sensing sensors, fusion of the features from multiple sensors, such as hyperspectral and light detection and ranging (LiDAR) is an effective method for land cover mapping. In this paper, we propose a novel ensemble classifier to fuse hyperspectral and LiDAR datasets for classification. First, morphological features are used to model spatial and elevation information from the first few principal components (PCs) of the original hyperspetcral (HS) image and LiDAR data. Second, we split different kinds of features (i.e., spectral bands, morphological features of hyperspectral and LiDAR), into several disjoint subsets and apply the data transformation method to each subset. In particular, three data transformation methods, including principal component analysis (PCA), linearity preserving projection (LPP) and unsupervised graph fusion (UGF) are considered. Third, the features extracted in each subset are concatenated to classify by a random forest (RF) classifier. Experimental results on a co-registered HS and LiDAR data provide the effectiveness and potentialities of the proposed ensemble classifier.

In this work, we develop a new framework to combine ensemble learning and composite kernel learning for hyperspectral image classification. We refer it as the multiple composite kernel learning, which is based on an iterative architecture. More specifically, in each iteration, we use the rotation-based ensemble to create rotation matrix, which is used to generate rotated features for both spectral and spatial information (e.g., extinction profiles). Then, the new spectral and spatial features are integrated into the composite kernels based on support vector machines classifier. Different rotation matrices will lead to obtaining various newly spectral and spatial characteristics, thereby they further increase the diversity and the classification performance. Experimental results on Indian Pines benchmark hyperspectral dataset demonstrate the excellent performance of the proposed method.

Advanced spaceborne thermal emission radiometer (ASTER) has been used to obtain global digital elevation models (DEMs) that cover all over the world using along-track stereovision; however, the coupling of nadir- and backward-observation is not free from some occlusion areas. Moreover, DEMs derived from optical stereovision suffer from voids due to clouds and shadows. Therefore, stereo pairs obtained from different cross-track pointing angles are investigated to obtain DEMs, which is called cross-track stereovision. Although DEMs by cross-track stereovision is produced by images acquired on different dates, the quantitative agreement is obtained, which shows that the cross-track stereovision is promising for DEM generation to fill the voids in global DEMs. Furthermore, DEMs obtained by cross-track thermal stereovision of ASTER are compared with that by along-track stereovision. Thermal infrared radiometer has an advantage of radiation measurement, which enables nighttime observation. These results show that further improvement in DEMs derived from ASTER is possible by increasing the number of possible stereo pairs.

Hyperspectral Imager Suite (HISUI) is a future spaceborne hyperspectral Earth imaging system being developed by Japanese Ministry of Economy, Trade, and Industry (METI). HISUI will be launched and deployed on International Space Station (ISS) for three year operation from 2019. In FY2016, manufacturing and testing of HISUI Flight Model and design of HISUI Exposed Payload System were conducted. HISUI data policy and research announcement as well as synergy with other earth observing instruments onboard ISS are also being discussed.

In the hyperspectral remote sensing community, decision forests combine the predictions of multiple decision trees (DTs) to achieve better prediction performance. Two well-known and powerful decision forests are Random Forest (RF) and Rotation Forest (RoF). In this work, a novel decision forest, called Partial Least Square Forest (PLSF), is proposed. In the PLSF, we adapt PLS to obtain the components for the hyperplane splitting. Moreover, the projection bootstrap technique is used to retain the full spectral bands for the selection of split in the projected space. Experimental results on three hyperspectral datasets indicated the effectiveness of the proposed PLSF because it enhances the diversity and accuracy within the ensemble when compared to RF and RoF.

In this work, we address the problem of unsupervised domain transfer learning via an ensemble strategy in the context of classification between multiple hyperspectral images. The objective of domain adaption is to assign the label to an image of interest (the target image) using the labeled samples in the source image. The proposed method is based on the rotation-based ensemble and transfer component analysis (TCA). In this method, the feature space in both source and target image is divided into several disjoint feature subsets. Then, the features induced by the TCA technique in the source domain are used as the input space to a random forest (RF) classifier. Finally, the results achieved by each step are fused by a majority vote. We compare the proposed method, ensemble of TCA (E-TCA), with a regular RF and an RF with the reduced features by the TCA. Experiments on the real hyperspectral image acquired over a Japanese mixed forest show remarkable cross-image classification performances.

Multiple classifier systems or ensemble learning is an effective tool for providing accurate classification results of hyperspectral remote sensing images. Two well-known ensemble learning classifiers for hyperspectral data are random forest (RF) and rotation forest (RoF). In this paper, we proposed to use a novel decision tree (DT) ensemble method, namely, canonical correlation forest (CCF). More specifically, several individual canonical correlation trees (CCTs) that are binary DTs, which use canonical correlation components for the hyperplane splitting, are used to construct the CCF. Additionally, we adopt the projection bootstrap technique in CCF, in which the full spectral bands are retained for split selection in the projected space. The techniques aforementioned allow the CCF to improve the accuracy of member classifiers and diversity within the ensemble. Furthermore, the CCF is extended to the spectral-spatial frameworks that incorporate Markov random fields, extended multiattribute profiles (EMAPs), and the ensemble of independent component analysis and rolling guidance filter (E-ICA-RGF). Experimental results on six hyperspectral data sets are used to indicate the comparative effectiveness of the proposed method, in terms of accuracy and computational complexity, compared with RF and RoF, and it turns out that CCF is a promising approach for hyperspectral image classification not only with spectral information but also in the spectral-spatial frameworks.

Hyperspectral Imager Suite (HISUI) is a future spaceborne hyperspectral Earth imaging system being developed by Japanese Ministry of Economy, Trade, and Industry (METI). HISUI project is currently being promoted by three organizations each of which has a contract with METI together with several scientists from universities and national research institutes. Recently, it was decided to deploy HISUI on International Space Station. The current status of HISUI project will be introduced in the presentation.

One of the fundamental issues in the generation of digital surface models (DSMs) from optical stereo imageries is the noise reduction on the low correlation areas. We apply the non-local means filtering to the DSM derived from Panchromatic Remote-sensing Instrument for Stereo Mapping (PRISM) onboard Advanced Land Observing Satellite (ALOS) with modifying the original algorithm so that it can appropriately remove the noise while preserving the original resolutions/profiles of the terrain. The results of the filtering are evaluated quantitatively with a reference DSM derived from an airborne LiDAR instrument, as well as qualitatively with visual inspections. The proposed algorithm shows the ability to improve global DSM datasets generated from ALOS PRISM.

Forest type classification using remote sensing data is affected by the limited number of training samples for the classifier owing to the cost of field surveys in practical situations. Although, hyperspectral data can provide rich spectral information, there is a small-sample-size problem with high-dimensional data owing to the limitation number of field samples; thus, the classification model has the risk of poor performance due to model overfitting. Therefore, band selection to avoid overfitting considering the characteristics of the forest environment are required to solve these problems. In this work, we propose a methodology based on sparse discriminant analysis (SDA) using hyperspectral data. It is found that our proposed methodology increased the classification accuracy under the small-sample-size conditions considering practical forest monitoring. Additional training data based on a tree crown density map calculated from very high resolution data further improved classification accuracy of SDA.

Since SWIR hyperspectral data provides useful information on minerals and soils, the applications to geology, mining and agriculture are expected. In the SWIR region, hyperspectral sensors use mercury cadmium telluride detectors that have more defects and show non-uniform responses to input radiances. Therefore, the pushbroom scanning of hyperspectral sensors causes stripe noises that disturb the minute spectral analysis. This work correct the artifact due to stripe noise at first using the conventional correction method, such as momentum matching. The remaining stripe patterns are corrected using a sparse coding technique, which is examined using real data.

We present a method for object detection based on sparse representations and Hough voting, which integrates sparse representations for local-feature detection into generalized-Hough-transform object detection. Object parts are detected via class-specific sparse image representations of patches using learned target and background dictionaries, and their cooccurrence is spatially integrated by Hough voting, which enables object detection. In this paper, a discriminative criterion is introduced into dictionary construction to improve the detection performance. Experiments performed on airplane detection and the identification of a specific ship show that the proposed method achieves state-of-the-art performance with the robustness against noise and occlusion using a small set of positive training samples.

Super-resolution framework is applied to airplane and satellite imagery to enhance the spatial property of multiple acquisition images. By using the brightness control, the differences caused by spectral and geometric configuration are adjusted. Subpixel shift is accurately obtained by phase correlation technique. Blur properties caused by pushbroom scanning systems are modeled using anisotropic functions. Maximum a posterior (MAP) solution is used to estimate high resolution images. Sequential airplane photo and ASTER imagery are used to apply the proposed methodology. The validity of the model is shown by checking the visual properties of reconstructed images, which is clearly observed on the edge response of buildings and piers.

Fusion of spectral and spatial information has good potential for building highly accurate classification model for land cover and prediction model for biomass estimation. In this study, a new method with spectral-spatial this fusion and object-based segmentation for monitoring peat swamp forest condition is proposed. Peatland is a major CO2emission source by peat burn, peat decomposition and forest fire. Remote sensing is effective tool for monitoring environmental condition of peatland and forest ecosystem. For the monitoring, forest type classification map and biomass distribution map are useful for understanding about the forest condition and to estimate mass volume of CO2storage. In order to have enough accurate maps without overfitting problem, sparse discrimination analysis (SDA) was applied to spectral and spatial information from hyperspectral data for the classification model, and LASSO regression was applied for the biomass prediction model. Furthermore, to obtain the well-segmented maps, mean shift clustering as object-based segmentation was applied to those maps for identifying suitable class and biomass with majority voting in each segmentation. These proposed scheme improved classification and prediction accuracies and provides accurate segmented maps.

Hyperspectral Imager Suite (HISUI) is a next-generation Japanese sensor that is planned to be on board in 2017 or later. HISUI is composed of two sensors; one is a hyperspectral sensor, and the other is a multispectral sensor. Synergic operation of these instruments acts an important role in the future spectral-spatial data. Regarding with Hyperspectral sensor, the onboard data correction and compression system are adopted in order to reduce the amount of downlink data as well as to realize the fine accuracy of radiometric and spectral performance after ground data processing. The investigation based on the data of the evaluation model of the instrument shows that the performance of onboard data correction can meet the requirement.

Digital Elevation Models (DEMs) and ortho-rectified images in Borneo (Kalimantan) island, where the probability of cloud coverage is too high to make DEMs by optical stereovision, are produced using ASTER data. By referring to Space Shuttle Radar Topography Mission (SRTM) data as the initial guess, we produced DEMs and orthoimages that make the most of ASTER data acquisition based on statistical estimation. We succeeded in making DEM in N00E114 area that utilizes ASTER data at 90 % of the total pixels, whereas that of ASTER Global DEM (GDEM) is composed of 60 % of ASTER data.

In this paper, we propose a new object localization method named sparse representation based object localization (SROL), which is based on the generalized Hough-transform-based approach using sparse representations for parts detection. The proposed method was applied to car and ship detection in remote sensing images and its performance was compared to those of state-of-the-art methods. Experimental results showed that the SROL algorithm can accurately localize categorical objects or a specific object using a small size of training data.

This paper presents an airborne experiment on unmixing-based hyperspectral super-resolution using RGB imagery. Preprocessing is described to ensure spatial and spectral consistency between hyperspectral and RGB images. An extended version of coupled nonnegative matrix factorization (CNMF) is introduced for multisensor hyperspectral super-resolution to deal with a challenging problem setting, i.e., only three spectral channels for higher spatial information and a 10-fold difference of ground sampling distance. The proposed method successfully estimated the high-spatial-resolution red-edge image. Numerical evaluation by comparing the high-spatial-resolution hyperspectral image to ground-measured spectra demonstrated recovery of pure-pixel spectra by the proposed method.

Nonlinear spectral mixture models have recently received particular attention in hyperspectral image processing. In this paper, we present a novel optimization method of nonlinear unmixing based on a generalized bilinear model (GBM), which considers the second-order scattering of photons in a spectral mixture model. Semi-nonnegative matrix factorization (semi-NMF) is used for the optimization to process a whole image in matrix form. When endmember spectra are given, the optimization of abundance and interaction abundance fractions converge to a local optimum by alternating update rules with simple implementation. The proposed method is evaluated using synthetic datasets considering its robustness for the accuracy of endmember extraction and spectral complexity, and shows smaller errors in abundance fractions rather than conventional methods. GBM-based unmixing using semi-NMF is applied to the analysis of an airborne hyperspectral image taken over an agricultural field with many endmembers, and it visualizes the impact of a nonlinear interaction on abundance maps at reasonable computational cost.

In hyperspectral remote sensing, a lot of endmember extraction algorithms have been proposed to deal with mixed-pixel problem. These algorithms are based on linear mixture models and search vertex points in the spectral space. However, practically it is unclear that endmember spectra lie around vertex positions in it. This paper presents a new endmember extraction approach that enhances endmember extraction algorithms. In our method, endmembers are clearly put into vertex positions after feature extraction based on continuum removal. This approach is applied to hyperspectral image (HSI) acquired by the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) over the Cuprite mining site in Nevada. An experiment shows that continuum removal makes the structure of the data cloud understandable even in low dimensional space, which results in enhancement of the SISAL, one of the endmember extraction algorithms.

Hyperspectral Imager Suite (HISUI) is a next-generation Japanese optical sensor that is composed of hyperspectral and multispectral imagers. Level 1 data processing is prepared to provide radiometrically and geometrically corrected data that act as standard data products of HISUI. Parallax correction is a key point of Level-1 data processing. Level 1G data product is orthorectified top-of-atmosphere radiance and is used as inputs for higher level data products. Atmospheric correction is one of higher level data products, which provides reflectance information.

Hyperspectral Imager Suite (HISUI) is a future spaceborne hyperspectral and multispectral Earth imaging system being developed by Japanese Ministry of Economy, Trade, and Industry (METI). HISUI project is currently being promoted by three organizations each of which has a contract with METI together with several scientists from universities and national research institutes. The current status of HISUI project will be introduced in the presentation.

A micro-satellite named “HODOYOSI-1” is a 50-kg-weight small remote sensing satellite. It has a pushbroom imaging system in multi-spectral-band: Red (R), Green (G), Blue (B), and two Near infrared (NIR). It is expected that, using super-resolution framework, high-resolution images are obtained from images acquired by two NIR sensors. In the process of super-resolution, it is necessary to estimate sub-pixel shifts and blur kernels of input images. Although pushbroom systems have different blur properties from normal camera systems, it has always been ignored in super-resolution for satellite images. To improve the results of super-resolution applied to pushbroom images, a new blur model, which is more suitable for pushbroom systems, are proposed. The validity of the new blur model is shown by computer simulations and experiments.

Hyper-spectral Imager Suite (HISUI) is a next-generation Japanese sensor that will be on board in 2016 or later. HISUI is composed of two radiometers; one is a hyperspectral imager that obtains spectral images of 185 bands with the ground sampling distance 30 meters from visible to shortwave-infrared region, and the other is a multispectral imager that covers the wide swath of 90 km with the ground sampling distance of 5 meters[1]. The sensor system is the follow-on mission of the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) in the visible and shortwave-infrared regions [2]. Synergic operation of these instruments acts an important role in the future spectral-spatial data. Simultaneously, the key components of the flight model are being developed. The design of flight model reflected from the test result of evaluation model as well as the development status of flight model is reported in this work.

Hyperspectral imager suite (HISUI) is the Japanese next-generation earth-observing sensor composed of hyperspectral and multispectral imagers. Unmixing-based fusion of hyperspectral and multispectral data enables the production of high-spatial-resolution hyperspectral data. HISUI simulated imaging system combining two imagers was developed for verification experiments to investigate the feasibility and clarify the whole procedure of the hyperspectral and multispectral data fusion mission on HISUI. Airborne experiments are planned as simulation tests of HISUI higher-order products. The experimental results of the ground based observation showed the importance of the preprocessing and cross-calibration on the final quality of fused data, which contributes to the practical use of hyperspectral and multispectral data fusion.

Image matching methods are integrated into digital elevation model (DEM) generation and some other satellite image processing, such as parallax correction. These applications require highly accurate disparity. In particular, a kind of DEM generation system called narrow baseline stereovision system requires fine sub-pixel image matching method. This system is attractive because of its ability to avoid an occlusion problem. In this work, to improve the estimation accuracy, we extended phase correlation technique to multi-band image. This extension can ensure robustness to target scenes, so that global accuracy of estimated disparity can be improved. We also applied proposed image matching method to commercial stereo camera images, which is useful to reduce miss-matching area compared with existing method.

In the hyperspectral era, the demand for data registration in spectral region is an important issue in addition to spatial region. Detection of smile and keystone phenomena that are caused by aberrations in spectrometer is related to registration activity, which is crucial for data fusion research. Hyperspectral Imager Suite (HISUI) is a next-generation Japanese optical sensor that is composed of a hyperspectral imager and a multispectral imager, which will be launched on Advanced Land Observation Satellite 3 (ALOS-3). Three similarity measures, normalized cross correlation (NCC), phase correlation (PC) and mutual information (MI), for spatial-spectral registration of hyperspectral data are discussed for Level-1 data processing of HISUI.

Hyperspectral Imager Suite (HISUI) is a next-generation Japanese optical sensor that is composed of a hyperspectral imager and a multispectral imager. Combination of hyperspectral and multispectral imagers contributes land remote sensing with high spectral and spatial resolution. Validation of the HISUI evaluation model, including environmental test such as vibration test and thermal-vacuum test has been carried out to meet the science requirements on the sensor under the space-borne conditions. Major performances of the sensors are obtained. Spectral response of each band of the hyperspectral sensor, distortions in a spectrometer in terms of smile and keystone properties, are investigated. The test results of the evaluation model show appropriate performance of the sensors.

Nonlinear spectral mixing models have recently been receiving attention in hyperspectral image processing. This work presents a novel optimization method for nonlinear unmixing based on a generalized bilinear model (GBM), which considers second-order scattering effects. Semi-nonnegative matrix factorization is used for optimization to process a whole image in a matrix form. The proposed method is applied to an airborne hyperspectral image with many endmembers and shows good performance both in unmixing quality and computational cost with simple implementation. The effect of endmember extraction on nonlinear unmixing is investigated and the impact of the nonlinearity on abundance maps is demonstrated.

At the core of the technical methodology for creating the Advanced Spaceborne Thermal Emission and Reflection radiometer (ASTER) global digital elevation model (GDEM) is the procedure for generating a global set of 1°latitude-by-1°longitude tiles containing DEM data in geographic latitude and longitude coordinates and with one arc second postings from scene-based ASTER DEMs. The ASTER GDEM is comprised of all tiles, which include at least 0.01% land in them, each containing 3601-by-3601 elevation data points. The tiles are created by stacking all observed scene DEM data matched geographically to the tile container, selecting valid data for each pixel, removing abnormal data values, and then averaging the remaining selected valid data to assign as the tile elevation data. Valid Earth surface elevation values typically clump within a ± 40-m range and are assumed to be lower in elevation than residual cloud outliers. The filtering process, which assigns the tile elevation data, is one of the most important parts of the GDEM generation system. The median-based selection method is designed to efficiently select the valid data for each pixel. The combination of cloud-masked and non-cloud-masked data is another important part of the process to assign accurate elevation data for each pixel, because the cloud masking capability is not perfect. The algorithm used to combine both data is described. The postprocessing for inland water bodies is successfully carried out to yield a flattened elevation value. This postprocessing is essential to assign unique elevation values for each inland water body. GDEM tile elevation data include some residual anomalies, mostly in areas with fewer than three valid stacked input scenes. The correction method using existing reference data also is described.

Coupled nonnegative matrix factorization (CNMF) unmixing is proposed for the fusion of low-spatial-resolution hyperspectral and high-spatial-resolution multispectral data to produce fused data with high spatial and spectral resolutions. Both hyperspectral and multispectral data are alternately unmixed into end member and abundance matrices by the CNMF algorithm based on a linear spectral mixture model. Sensor observation models that relate the two data are built into the initialization matrix of each NMF unmixing procedure. This algorithm is physically straightforward and easy to implement owing to its simple update rules. Simulations with various image data sets demonstrate that the CNMF algorithm can produce high-quality fused data both in terms of spatial and spectral domains, which contributes to the accurate identification and classification of materials observed at a high spatial resolution.

Hyperspectral Imager Suite (HISUI) is a next-generation Japanese optical sensor that is composed of a hyperspectral imager and a multispectral imager, which will be launched on Advanced Land Observation Satellite 3 (ALOS-3). Combination of hyperspectral and multispectral imagers contributes land remote sensing with high spectral and spatial resolution. Calibration of the HISUI functional-equivalent model has been carried out to meet the science requirements on the sensor. A center wavelength and spectral response of each band of the hyperspectral sensor is obtained, from which distortions in a spectrometer in terms of smile and keystone properties are investigated. The results of the calibration activity for the functional-equivalent model show appropriate performance of the sensors.

Registration between multimodal images is carried out using mutual information (MI) as a similarity measure. MI registration does not require linearity between two images; therefore, this method is effective in registration between multimodal images. First, corresponding points between multiband images are obtained in sub-pixel level. We evaluated the sub-pixel registration errors between two images. In this case, MI shows more accurate performance than sum square distance (SSD) and normalized cross-correlation (NCC). This result shows that the technique is robust with respect to variants of illumination or wavelength of observation band. Second, image-to-map registration is also successful. These results proved the potential of MI to build GIS with satellite images and map data.

Hyperspectral sensors (HS) are next-generation optical sensors that have excellent spectroscopic performance with hundreds of bands. Multispectral sensors (MS) are conventional optical sensors that have several bands with higher spatial resolution. There is a trade-off relation between spectral and spatial performance of optical sensors in terms of signal-to-noise ratio and data rate. In order to obtain high resolution and spectral information simultaneously, pan-sharpen approach is applied to fuse them. Hyperion as a hyperspectral sensor, and ASTER as a multispectral sensor, are chosen in this work. In addition, ideal simulation of fusion is carried out using AVIRIS image. For the purpose of quantitative evaluation, five well-known quality indices are used. Excellent result of newly proposed methods based on pan-sharpen approach is confirmed by both of ideal simulation using AVIRIS and Hyperion-ASTER fusion.

The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) has acquired more than 1 million data with along-track stereovision. These data are processed to make digital elevation model (DEM) with orthorectified images. Using multiple DEMs obtained at different times, it is expected to enhance the spatial resolution. In this work, the methodology to enhance the resolution of DEM, especially Bayesian methodology is discussed.

The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) Global Digital Elevation Model (GDEM) was released to the public in My 2009. The GDEM version 2, currently under development, is reproduced using an updated algorithm and contains new ASTER data observed after September 2008. Validation study of the trial and beta version of GDEM version 2 was carried out to investigate its characteristics such as geolocation error, elevation error and horizontal resolution. The results of the study confirmed that elevation offset and horizontal resolution will be greatly improved in version 2 and the enhanced horizontal resolution will serve to reduce the standard deviation of elevation and geolocation error.

Coupled non-negative matrix factorization (CNMF) is introduced for hyperspectral and multispectral data fusion. The CNMF fused data have little spectral distortion while enhancing spatial resolution of all hyperspectral band images owing to its unmixing based algorithm. CNMF is applied to the synthetic dataset generated from real airborne hyperspectral data taken over pasture area. The spectral quality of fused data is evaluated by the classification accuracy of pasture types. The experiment result shows that CNMF enables accurate identification and classification of observed materials at fine spatial resolution.

Image matching and water-body-detection methodologies are essential parts of generating good-quality digital elevation model (DEM) data. It is one of the very important results for image matching where 1-D searching in the along-track direction is sufficient to find the maximum correlation point if reconstructed unprocessed Advanced Spaceborne Thermal Emission and Reflection data (Level-1A data) are used as the source data for DEM products. This important situation is obtained from the general formulation of how to make 1-D searching possible. The image matching quality is evaluated for this 1-D searching method. An image correlation kernel size of 5 by 5 is recommended as the most suitable selection for better horizontal resolution with a slight sacrifice of the image matching error. The satellite pointing fluctuation effect on image matching is also evaluated, leading to the conclusion that it does not seriously affect DEM quality. The water-body-detection technique is another core of DEM generation. The low image correlation coefficient, the low reflectance of water in the near-infrared band 3N, and other spectral characteristics of water were used to identify surface water bodies. In addition, water-body size and the standard deviation of the water-body perimeter elevation are limited for consistent detection without misidentification. As a result, the minimum size of a detectable water body is 0.2km2.

We describe a methodology that allows the more accurate detection and investigation of 3-D terrain changes using digital elevation models (DEMs) derived from stereoscopic pushbroom cameras on board satellites. Repeat 15-m-resolution orthorectified images and stereophotogrammetric DEMs were produced from the Advanced Spaceborne Thermal Emission and Reflection Radiometer for displacement analysis. The column and line numbers in the sensor geometry are included to each pixel. We reveal that the attitude oscillation of the satellite has a considerable effect on geometric accuracy, which is, in this paper, corrected along the scan lines of pushbroom sensors by destriping in the sensor geometry. Our results show that the distortion in DEMs is reduced by about 40% in terms of standard deviation under similar sun elevation conditions by correcting the effect of pitch and yaw oscillations. The proposed methodology was applied to the measurement of ground deformation generated by a swarm of earthquakes between September 14 and October 4, 2005, which occurred in Dabbahu, Ethiopia. We successfully detected the horizontal displacements and DEM differences with about 0.1-pixel accuracy.

Since the visible detector of the Multiband Imager (MI) of the Lunar Imager/Spectrometer mounted on the Selenological and Engineering Explorer operates in a frame-transfer mode of a 2-D charge-coupled device, signals suffer from additional light exposure during the charge transfer, causing complicated artifacts. In this paper, we developed an iterative image-correction algorithm based on a model of multiband artifact phenomena. Using lunar images acquired while in orbit, we validated the correction performance based on a cross-comparison with near-infrared images of the MI obtained at a similar wavelength. Our correction algorithm corrected the additional electric charge of up to 10% of the obtained signal at the Apollo 16 site. Photometric and terrain corrections that include the effect of the observation angle were found to be important in precise cross-calibration.

Accuracy of digital elevation model (DEM) obtained by the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) that has along-track stereovision is investigated. The pointing offset and stability of the radiometer is one cause of the geometric deviation of the ASTER DEM attached with orthorectified image. The correction methodology to be implemented to the data processing is suggested. A fine-tuning of image matching procedure leads to better reproduction of the topography. The comparison with reference DEM is described.

The hyperspectral and the multispectral (hyper-multi spectral) mission is the Japanese next generation space-borne radiometer development project. This project is a heritage from ASTER launched in December 1999. The performance of the hyperspectral radiometer is 30m ground sampling distance, 30km swath width, 10nm and 12.5nm wavelength distance for VNIR and SWIR respectively, over 450@620nm and 300@2,100nm of the signal to noise (S/N) ratio. The performance of the multispectral radiometer is 5m ground sampling distance, 90km swath width, over 200 for all bands of S/N ratio. This project will be launched on ALOS-3 of JAXA in FY2014. The panchromatic sensor with stereo viewing will also be installed on ALOS-3.

In this study, we demonstrate an automatic methodology for estimating the satellite pitch vibration by using information of sensors with small parallax and correcting the images taken by the sensors themselves. Consequently, We can get corrected digital elevation model (DEM) from the images disturbed by satellite pitch vibration. A satellite often vibrates in roll and pitch directions, and each component has a bad influence on photogrammetry. Since the pitch component directly deteriorates the accuracy of DEM in case of along-track stereo sensor, we focus on estimating and correcting pitch vibration. Our methodology is particularly useful for small parallax stereo system, because sub-pixel image matching precision is needed for small parallax stereo system and satellite vibration causes a critical problem.

Hyperspectral imaging sensors suffer from spectral and spatial misregistrations. These artifacts prevent the accurate acquisition of the spectra and thus reduce classification accuracy. The main objective of this work is to detect and correct spectral and spatial misregistrations of hyperspectral images. The Hyperion visible near-infrared (VNIR) subsystem is used as an example. An image registration method using normalized cross-correlation for characteristic lines in spectrum image demonstrates its effectiveness for detection of the spectral and spatial misregistrations. Cubic spline interpolation using estimated properties makes it possible to modify the spectral signatures. The accuracy of the proposed postlaunch estimation of the Hyperion properties has been proven to be comparable to that of the prelaunch measurements, which enables the precise onboard calibration of hyperspectral sensors.

Accurate attitude estimation of spacecrafts is the main requirement to provide good geometric performance of remote-sensing imagery. The Advanced Spaceborne Thermal Emission and Reflection Radiometer/short-wave-infrared subsystem has six linear-array sensors arranged in parallel, and each line scans the same ground target with a time interval of 356.238 ms between neighboring bands. The registration performance between bands becomes worse when attitude fluctuation occurs during a time lag between observations. Since the time resolution of the line scan is higher than that of the attitude information provided from the satellite, attitude data are estimated with a high frequency. We succeeded in correcting the image-registration error using the revised attitude information. As a result, the image distortion of 0.2 pixels caused by spacecraft-attitude jitter is reduced to less than 0.08 pixels, showing that band-to-band registration errors of a sensor with parallax observation are available to improve the image distortion caused by attitude fluctuation.

The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) system acquires multispectral images ranging from the visible to thermal infrared region. The ASTER system consists of three subsystems: visible and near-infrared (VNIR), short-wave infrared (SWIR) and thermal infrared (TIR) radiometers. The VNIR subsystem has a backward-viewing telescope as well as a nadir one. To deliver data products of high quality from the viewpoint of geolocation and band-to-band registration performance, a fundamental program, called Level-1 data processing, has been developed for images obtained using four telescopes with a cross-track pointing function. In this work, the methodology of the geometric validation is first described. Next, the image quality of ASTER data products is evaluated in view of the geometric performance over a period of four years. The band-to-band registration accuracy in the subsystem is better than 0.1 pixels and that between subsystems is better than 0.2 pixels. This means that the geometric database is determined accurately and the image matching method based on a cross-correlation function is effective in the operational usage.

Stray light components in images obtained by the shortwave infrared (SWIR) and visible near-infrared (VNIR) radiometers of the Advanced Spaceborne Thermal Emission Reflection Radiometer (ASTER) were investigated. A simple method, which is equivalent to the van Cittert method of deconvolution, was used for correction. The stray light components were estimated using the image obtained during lunar observation, and the improvement in image quality was examined after stray light correction. The calculation is performed in the space domain, and application to the filter scratch problem of the ASTER/SWIR sensor, which has a scratch on the interference filter resulting in partially degenerated images, is also demonstrated.

The mechanism of crosstalk phenomena in the shortwave infrared (SWIR) subsystem of the Advanced Spaceborne Thermal Emission and Reflection Radiometer, which has six bands in the wavelength region of 1.6-2.43 /spl mu/m, is investigated. It is found that light incident to band 4 is reflected at the detector and the filter boundary, and then transported to other bands by multiple reflections in the focal plane area. A crosstalk correction algorithm is developed to improve the spectral separation performance of SWIR. Parameters of the crosstalk model, i.e., the amount of stray light and its area of influence, are determined by image analysis. By careful investigation of SWIR images around peninsulas, lakes, and islands, the crosstalk model is validated. Therefore, the correction algorithm is implemented in the preprocessing of higher level data products.