This paper presents the results of the investigation on image fusion applied to data with significantly different spectral properties. The experiments with simulated and real data show that conventional fusion methods heavily distort geometrical shapes of features in case of fusion of data with different spectral characteristics. As a step forward to solve the problem, a novel fusion approach is proposed. The method uses the relationship between the Wavelet Transform Modulus Maxima (WTMM) corresponding to the same feature in the fusing images. The experiments showed advantages of the proposed method in areas of substantially dissimilar spectral characteristics of merging data. The found limitations of the method are discussed as well.

Compact High Resolution Imaging Spectrometer onboard the Project for On-board Autonomy, or CHRIS/Proba, represents a new generation of satellite images that provide different acquisitions of the same scene at five different angles. Given the hyperspectral-oriented waveband configuration of the CHRIS images, the scope of its application would be much wider if the present 17m nadir resolution could be refined. This paper presents the results of three superresolution methods applied to multiangular CHRIS/Proba data. The CHRIS images were preprocessed and then calibrated into reflectance using the method described in [1][2]. Automatic registration using an intensity variation approach described in [3] was implemented for motion estimation. Three methods, namely non-uniform interpolation and de-convolution [4], iterative back-projection [5], and total variation [6] are examined. Quantitative measures including peak signal to noise ratio [7], structural similarity [8], and edge stability [9], are used for the evaluation of the image quality. To further examine the benefit of multi-frame superresolution methods, a single-frame superresolution method of bicubic resampling was also applied. Our results show that a high resolution image derived from superresolution methods enhance spatial resolution and provides substantially more image details. The spectral profiles of selected land covers before and after the application of superresolution show negligible differences, hinting the use of superresolution algorithm would not degrade the capability of the data set for classification. Among the three methods, total variation gives the best performance in all quantitative measures. Visual inspections find good results with total variation and iterative back-projection approaches. The use of superresolution algorithms, however, is complex as there are many parameters. In this paper, most of the parameter settings were tuned manually or decided empirically.

The MTF (modulation transfer function) has been used to evaluate the performance of optical systems. More accurate access to on-orbit MTF opens the way to enhance images so that we can deduce more information from the images of IKONOS, OrbView, or SPOT. Generally, the MTF was assessed using various methods such as the point source method and the knife-edge method etc. In this paper, the value of the slanted-edge method is investigated as a tool to assess the on-orbit MTF. The slanted-edge method is the ISO 12233 standard for the MTF measurement of electronic still-picture cameras. The method is adapted to estimate the MTF values of line-scanning telescopes. We applied the method to measure the MTF of high resolution satellite images and also, we compared the MTF measurement with that of the knife-edge method applied to EOS-C test images to increase the confidence on MTF measurement on EOS-C. EOS-C is a 2.5m resolution satellite sensor for SI-200 satellite which is under the development by Satrec Initiative. The MTF measured from this method is used for the MTF compensation of satellite images by generating the MTF convolution kernel based on the PSF (point spread function).

In this paper, mixed spectral-structural kernel machines are proposed for the classification of very-high resolution images. The simultaneous use of multispectral and structural features (computed using morphological filters) allows a significant increase in classification accuracy of remote sensing images. Subsequently, weighted summation kernel support vector machines are proposed and applied in order to take into account the multiscale nature of the scene considered. Such classifiers use the Mercer property of kernel matrices to compute a new kernel matrix accounting simultaneously for two scale parameters. Tests on a Zurich QuickBird image show the relevance of the proposed method : using the mixed spectral-structural features, the classification accuracy increases of about 5%, achieving a Kappa index of 0.97. The multikernel approach proposed provide an overall accuracy of 98.90% with related Kappa index of 0.985.

The May 2006 Central Java earthquake (M_w=6,3) heavily struck the Special Province of Yogyakarta, Indonesia. The widespread damage and high number of casualties revealed the necessity of vulnerability assessment within this region. A vulnerability and risk analysis was conducted using a GIS-based Indicator-Index Method. In this analysis very high resolution satellite images were used for two purposes: Firstly to validate the risk map by detecting damaged areas. This lead to the development of an object-based and a pixel-based methodology for damage detection. The detected damage areas were analyzed on a building unit scale. For each building unit the damage was calculated. Secondly to extract detailed information on the spatial distribution of landuse in the study area. This was conducted using a object-based, multiresolution approach.

Identification of land cover types is one of the most critical activities in remote sensing. Nowadays, managing land resources by using remote sensing techniques is becoming a common procedure to speed up the process while reducing costs. However, data analysis procedures should satisfy the accuracy figures demanded by institutions and governments for further administrative actions. This paper presents a methodological scheme to update the citrus Geographical Information Systems (GIS) of the Comunidad Valenciana autonomous region (Spain). The proposed approach introduces a multi-stage automatic scheme to reduce visual photointerpretation and ground validation tasks. First, an object-oriented feature extraction process is carried out for each cadastral parcel from very high spatial resolution (VHR) images (0.5m) acquired in the visible and near infrared. Next, several automatic classifiers (decision trees, multilayer perceptron, and support vector machines) are trained and combined to improve the final accuracy of the results. The proposed strategy fulfills the high accuracy demanded by policy makers by means of combining automatic classification methods with visual photointerpretation available resources. A level of confidence based on the agreement between classifiers allows us an effective management by fixing the quantity of parcels to be reviewed. The proposed methodology can be applied to similar problems and applications.

This paper deals with a comparison of different colour space in order to improve high resolution images classification. The background of this study is the measure of the agriculture impact on the environment in islander context. Biodiversity is particularly sensitive and relevant in such areas and the follow-up of the forest front is a way to ensure its preservation. Very high resolution satellite images are used such as QuickBird and IKONOS scenes. In order to segment the images into forest and agriculture areas, we characterize both ground covers with colour and texture features. A classical unsupervised classifier is then used to obtain labelled areas. As features are computed on coloured images, we can wonder if the colour space choice is relevant. This study has been made considering more than fourteen colour spaces (RGB, YUV, Lab, YIQ, YC_rC_s, XYZ, CMY, LMS, HSL, KLT, IHS, I₁I₂I₃, HSV, HSI, etc.) and shows the visual and quantitative superiority of IHS on all others. For conciseness reasons, results only show RGB, I₁I₂I₃ and IHS colour spaces.

This paper presents a semi-supervised one-class support vector machine classifier for remote sensing applications. In one-class image classification, one tries to detect pixels belonging to one class and reject the others. When few labeled target pixels and no labeled outlier pixels are available, the selection of the support vector machine free parameters is very challenging. This problem can be alleviated by introducing the information of the wealth of unlabeled samples present in the scene. The proposed algorithm deforms the training kernel by modelling the data marginal distribution with the graph Laplacian built with labeled and unlabeled samples. The good performance of the proposed method is illustrated in challenging remote sensing image classification scenarios where information of only one class of interest is available. In particular, we present results in multispectral cloud screening, hyperspectral crop detection, and multisource urban monitoring. Experimental results show the suitability of the proposal, specially in cases with few or poorly representative labeled samples.

In this paper we present a system for the estimation of forest biomass at individual tree level, which is based on multireturn LIDAR data and on Support Vector Regression (SVR). In particular, we estimate stem diameters that, combined with tree heights (provided by the Digital Canopy Model extracted from LIDAR data), allow one to retrieve tree biomass. The system proposed is made up of a preprocessing module, a LIDAR segmentation algorithm (aimed at retrieving tree crowns), a variable extraction and selection procedure and an estimation technique. The variables derived from LIDAR data, which are computed from both the intensity and elevation channels of all available returns, are representative of the characteristics of tree crowns. Four different methods of variable selection have been analyzed, and the sets of variables obtained have been used in the estimation phase based on the Support Vector Regression (SVR) technique. Experimental results show that the system proposed is effective for the estimation of stem diameters and tree biomass.

The Orthogonal Subspace Projection (OSP) algorithm is substantially a kind of matched filter that requires the evaluation of a prototype for each class to be detected. The kernel OSP (KOSP) has recently demonstrated improved results for target detection in hyperspectral images. The use of kernel methods helps to combat the high dimensionality problem and makes the method robust to noise. This paper incorporates the contextual information to KOSP with a family of composite kernels of tunable complexity. The good performance of the proposed methods is illustrated in hyperspectral image target detection problems. The information contained in the kernel and the induced kernel mappings is analyzed, and bounds on generalization performance are given.

The concept of virtual dimensionality (VD) has been developed for estimating the number of spectrally distinctive signals in a hyperspectral image. It has important applications in hyperspectral image analysis. For instance, it is related to the number of classes in classification and the number of endmembers in linear mixture analysis; an appropriate VD estimate will facilitate the related algorithm implementation and improve their performance. In this paper, we will evaluate several VD estimation approaches, including a Neyman-Pearson Detection based method and a Signal Subspace Estimation based method. In particular, we will discuss how the noise estimation affects the accuracy of VD estimate.

Hyperspectral sensors are being developed for remote sensing applications. These sensors produce huge data volumes which require faster processing and analysis tools. Vertex component analysis (VCA) has become a very useful tool to unmix hyperspectral data. It has been successfully used to determine endmembers and unmix large hyperspectral data sets without the use of any a priori knowledge of the constituent spectra. Compared with other geometric-based approaches VCA is an efficient method from the computational point of view. In this paper we introduce new developments for VCA: 1) a new signal subspace identification method (HySime) is applied to infer the signal subspace where the data set live. This step also infers the number of endmembers present in the data set; 2) after the projection of the data set onto the signal subspace, the algorithm iteratively projects the data set onto several directions orthogonal to the subspace spanned by the endmembers already determined. The new endmember signature corresponds to these extreme of the projections. The capability of VCA to unmix large hyperspectral scenes (real or simulated), with low computational complexity, is also illustrated.

In this paper, the original N-FINDR algorithm is implemented in four different versions: 1) parallel mode using the data after dimensionality reduction; 2) parallel mode using all the original bands; 3) sequential mode using the data after dimensionality reduction; 4) sequential mode using all the original bands. We will analyze the performance discrepancy among these versions. Based on experimental evaluation and comparison, instructive recommendations in implementation strategy for practical applications are provided.

In surveillance and remote sensing applications images are often subject to processing both at the sensor and for display/storage. This paper presents a step towards measuring and understanding how these processes impact on human interpretation of regions of similar statistics within images. The paper describes the methods involved, including image generation, image processing algorithm application, and algorithm impact measurement techniques. A comparison is then made between the impact measurements and human observations. It is suggested that data on mathematical measures of the impact of processing algorithms on statistical regions of images may be usable for intelligent algorithm application.

In our laboratory, an 8 kg nano-scale remote sensing satellite "PRISM" has been developed, which has an extensible boom with a telephoto lens on the tip, being a refracting telescope in the orbit. Imaging using such flexible optical systems bears the problem of structural deformation which results in low quality of the captured images. To compensate for it, the satellite has to have a focusing function in limited techniques. This paper proposes a practical way of auto focusing for such nano-scale remote sensing satellites using fuzzy image processing. The proposed method is independent of photographed objects and does not require tracking of the contrast evaluation areas to be used for conventional auto focusing algorithm, which reduces the on-board computational load and requirement of attitude determination and control very much.

Automatic image registration is a process related to various application fields such as remote sensing, medicine, computer vision, among others. Particularly in remote sensing, the ever increasing number of available satellite images asks for automatic image registration methods, capable of correctly align a new image. An automatic image registration method is proposed, based on the identification of a thin line through the Hough transform, from the diagonal brighter strip visible on the correlation images. This procedure is applied for both directions. Dividing an image into tiles and taking the center of each tile as a point, a geometric correction at the subpixel level may be achieved. Measures for an objective assessment of the geometric correction quality are also proposed, as a complement to the traditional RMS of the errors and visual inspection. An orthorectified Landsat image and an ASTER image with an approximate geometric correction were used. The images were superimposed and resampled. An image registration with subpixel accuracy was achieved. The proposed methodology has showed to be able to correctly align two images, having a priori a "gold standard" image covering a considerable part of the image to be registered.

The subject of the present paper is how are the statistical parameters of an image of a simple band ratio are influenced, in quantitative terms, by the correlation coefficient between the two bands. The possibility to modify the expression for the ratio of highly correlated bands, in order to enhance the tonality contrast between the target of interest and the adjacent pixels, is also studied. Using proper bivariate Gaussian distributions it is shown that the standard deviation and the entropy of the image decrease considerably, as well as the correlation coefficient increases. When the correlation coefficient is high, it is possible to use the square of the simple ratio, in order to improve the contrast of the image. The results and conclusions of this paper may be useful in mapping the vegetation cover or detecting lithological units and mineralization zones with the aid of remote sensing technology.

Visual tracking is an important task that has received a lot of attention in recent years. Robust generic tracking tools are of major interest for applications ranging from surveillance and security to image guided surgery. In these applications, the objects of interest may be translated and scaled. We present here an algorithm that uses scaled normalized cross-correlation matching as the likelihood within the particle filtering framework. We do not need color and contour cues in our algorithm. Experimental results with constant rectangular templates show that the method is reliable for noisy and cluttered scenarios, and provides accurate and smooth trajectories in cases of target translation and scaling.

The PIRATE, short for Passive Infra Red Airborne Tracking Equipment, multi mode Infrared Sensor, has been developed and manufactured by the Eurofirst consortium. The consortium is led by SELEX Galileo, contract award since 1992, from Italy, and includes THALES UK, and TECNOBIT from Spain. At the end of the development phase, a three years flight test campaign has started in 2005, and has been carried out involving Eurofighter partner companies, as flight trials responsible, like ALENIA in Turin (FLIR flight testing leading company), BAE SYSTEM in Warton, and EADS in Manching, highlighting the functionality and the performance of the unit, when operating in real scenarios, with cooperative target and TOO (target of opportunity), across different weather conditions, like clean sky, but also misty and cloudy weather, haze and hail, generally not ideal to the infrared wavelength operation, either during tracking or imaging.

A two-stage method for segmenting and estimating the motion of ionospheric electron density enhancements is presented. These enhancements occur during geomagnetic storms, when matter ejected from the sun enters the upper atmosphere. The first stage of the proposed method segments the enhancements in the images by using a tuned attribute morphology filter, based on feature contrast. A temporal feedback mechanism is also employed to improve the temporal stability of the outputs. The second stage makes use of shape contexts to calculate boundary matches. These matches are then used to estimate motion vectors and a post filter is applied to detect and correct anomalous vectors. Overall, the scheme produces data-driven estimates of the motion of ionospheric electron density enhancements that can be used to study the evolution of ionospheric storms. Illustrative results are presented for a sequence of images from the Halloween storm of 2003. The results of this technique can also be used to validate models and data from other instrumentation which images the ionosphere during storms.

This paper presents a detailed experimental study on the behavior of the backscattering of buildings in very high resolution (VHR) synthetic aperture radar (SAR) images under varying conditions. The double bounce effect caused by the corner reflector between the front wall of the buildign and its surrounding ground area is an important characterist of the building in VHR SAR. Therefore, we focus on the analysis of the relation between the double bounce effect and the aspect angle of the building. The study is carried out in three phases: i)development of a laboratory experimental setup on a scaled building model under well-controlled conditions with a variety of viewing configurations; ii) validation of the results obtained from the laboratyr measurements with real VHR airborne SAR data; iii) comparison of the above mentioned results with the simulations obtained by two theoretical models derived from electromagnetic theory. The laboratory experiments were carried out at the European Microwave Signature Lab (EMSL) at the Joint Research Centre (JRC) of the European Commission (EC), whie the real airborne SAR images wer acquired by the RAMSES sensor and were processed in order to obtain simulations of COSMO-SkyMed satellite images. The analyses showed that the strength of the double bounce drops rapidly in the low aspect angle range, while it decreases moderately for larter angles.

A common assumption concerning noise in radar images is that it is of multiplicative nature and spatially uncorrelated. Meanwhile, recent studies have shown that additive noise component cannot be neglected, especially for images formed by side look aperture radars (SLARs). Moreover, majority of radar image filtering techniques are designed under assumption that noise is i.i.d., i.e. spatially uncorrelated. However, in many practical situations the latter assumption is not true. Besides, spatial correlation properties of noise can be different and they are often a priori unknown. In this paper we demonstrate that complex statistical and spatial correlation characteristics of noise in radar images can and should be taken into consideration at image filtering stage. We design a modification of the denoising algorithm based on discrete cosine transform (DCT) that is able to easily incorporate a priori information or obtained estimates of noise statistical and spatial correlation characteristics. This can be done in automatic (blind) manner due to utilizing a sequence of blind estimation operations. We present simulation results that show appropriate accuracy and robustness of these operations. Finally, real life image filtering examples are given that confirm the effectiveness of the designed techniques.

Multitemporal analysis of Synthetic Aperture Radar (SAR) images has gained an ever increasing attention due to the availability of several satellite platforms with different revisit times and to the intrinsic capability of the SAR system of producing all-weather observations. As a drawback, automated analysis in general and change detection in particular are made dfficult by the inherent noisiness of SAR imagery. Even if a preprocessing step aimed at speckle reduction is adopted, most of algorithms borrowed from computer vision cannot be profitably used. In this work, a novel pixel feature suitable for change analysis is derived from information-theoretic concepts. It does not require preliminary despeckling and is capable of providing accurate change maps from a couple of SAR images. The rationale is that the negative of logarithm of the probability of an amplitude level in one image conditional to the level of the same pixel in the other image conveys an information on the amount of change occurred between the two passes. Experimental results carried out on two couples of multitemporal SAR images demonstrate that the proposed IT feature outperforms the Log-Ratio in terms of capability of discriminating either burnt or flooded areas and is less sensitive than Log-Ratio to changes in acquisition angle between the two SAR images.

Principal component analysis (PCA) is often used to detect change over time in remotely sensed images. A commonly used technique consists of finding the projections along the two eigenvectors for data consisting of two variables which represent the same spectral band covering the same geographical region acquired at two different time points. If change over time does not dominate the scene, the projection of the original two bands onto the second eigenvector will show change over time. In this paper a kernel version of PCA is used to carry out the analysis. Unlike ordinary PCA, kernel PCA with a Gaussian kernel successfully finds the change observations in a case where nonlinearities are introduced artificially.

This work deals with a novel adaptive parcel-based method for change detection in long time series of SAR images. The proposed technique considers two temporal series of SAR images acquired over the same geographical area in different periods and is based on the following 3 steps: a) adaptive modeling of the geometry of multitemporal SAR sequences according to the generation of spatio-temporal homogeneous regions (i.e., spatio-temporal parcels); b) comparison of series according to a parcel-based similarity measure (e.g., the Kullback-Leibler distance); and c) change-detection map generation according to thresholding. Parcels result in a proper modeling of complex spatial-temporal phenomena in the scene as well as borders and details of the changed areas. The use of similarity measures between long temporal series allows one capturing the complexity of multitemporal data involving statistics of different order. Experiments carried out on two sequences of ERS-1/ERS-2 SAR data confirmed the effectiveness of the proposed approach.

In this paper we propose a lossless compression algorithm for hyperspectral images based on distributed source coding; this algorithm represents a significant improvement over our prior work on the same topic, and has been developed during a project funded by ESA-ESTEC. In particular, the algorithm achieves good compression performance with very low complexity; moreover, it also features a very good degree of error resilience. These features are obtained taking inspiration from distributed source coding, and particularly employing coset codes and CRC-based decoding. As the CRC can be used to decode blocks using a reference different from that used to compress the image, this yields error resilience. In particular, if a block is lost, decoding using the closest collocated block in the second previous band is successful about 70% of the times.

Remote sensing images are commonly formed on-board an observation platform, then transferred via a communication downlink, and finally processed on-land. There are many ways of compressing and then classifying remote sensing images. In this paper we focus on considering two lossy compression techniques under the assumption that the original images are noisy. No pre- or postprocessing is applied. Two classifiers are examined, namely, those based on trained radial basis function neural networks and support vector machines. We study how the parameter that controls the compression ratio of two coders based on the discrete cosine transform influences classification accuracy of these classifiers for a real life three-channel optical image. It is shown that attaining the optimal operation point for both coders is practically equivalent to providing the maximal probability of correct classification of multichannel data. At the same time, the efficiency of image compression characterized in terms of compression ratio, peak signal-to-noise ratio, and probability of correct classification considerably depends upon the coder used. Finally, it is shown that compressing multichannel remote sensing data in the neighborhood of the optimal operation point and near the maximum of the probability of correct classification can be performed in automatic manner.

Remote sensing has become an unavoidable tool for better managing our environment, generally by realizing maps of land cover using classification techniques. The classification process requires some pre-processing, especially for data size reduction. The most usual technique is Principal Component Analysis. Another approach consists in regarding each pixel of the multispectral image as a mixture of pure elements contained in the observed area. Using Blind Source Separation (BSS) methods, one can hope to unmix each pixel and to perform the recognition of the classes constituting the observed scene. Our contribution consists in using Non-negative Matrix Factorization (NMF) combined with sparse coding as a solution to BSS, in order to generate new images (which are at least partly separated images) using HRV SPOT images from Oran area (Algeria). These images are then used as inputs of a supervised classifier integrating textural information. The results of classifications of these "separated" images show a clear improvement (correct pixel classification rate improved by more than 20%) compared to classification of initial (i.e. non separated) images. These results show the contribution of NMF as an attractive pre-processing for classification of multispectral remote sensing imagery.

Traditional air pollution monitoring by using ground based instruments cannot provides air pollution over a large spatial scale. Aerosol over Penang Island was mapped with SPOT satellite image in this study. The objective of this study is to evaluate the relationship between SPOT satellite observation and particulate matter of size less than 2.5 micron (PM2.5) parameter. It is based on detection of dark surface targets in the blue band. Only one visible wavelength band were used in this study. The surface reflectance values for the only one visible wavelength were determined based on the information given by using ATCOR2 image processing technique. The atmospheric components were then estimated from the image. A total of 25 ground truth PM2.5 data were collected simultaneously with the acquired satellite image by using a hand held DustTrak Meter and their locations were determined using a hand held Global Positioning System (GPS). The digital numbers of the corresponding in situ data were converted into irradiance and then reflectance. The atmospheric reflectance values was extracted from the satellite observation reflectance values subtrated by the amount given by the surface reflectance. The atmospheric reflectance values were later used for PM2.5 mapping using the calibrated algorithm. An algorithm was developed based on the atmospheric optical characteristic. The developed algorithm was used to correlate the digital signal and the PM2.5 concentration. A good linear correlation between the satellite signal and PM2.5 parameter was found in this study (R > 0.8). Finally, a PM2.5 map was generated using the proposed algorithm. This study indicates that the feasibility of using the visible band from SPOT for PM2.5 mapping.

The image fusion technique is to maximize the information in images at same area or object taken by different sensors. It enhances unapparent features at each image and wildly applied at remote sensing, medical image, machine vision, and military identification. In remote sensing, the latest sensors usually provide 11-bit panchromatic data which compose more radiometric information; however the standard visual equipment can only produce 8-bit resolution content that limits the analysis of imagery on the screen or paper. This paper shows how to preserve the original 11-bit information after the DRA (Dynamic Range Adjustment) approaches and keep the output from color distortion during the following pan/multi-spectrum image fusion process. We propose a good dynamic range compression method converting the original IKONOS panchromatic image into high, low luminance and typical linear stretched images and using wavelet fusion to enhance the radiometric visualization and keeping good correlation with the multi-spectrum images in order to produce fine pan-sharpened product.

Waveform calibration is a crucial task in the processing of full-waveform laser scanner data. In most cases, there is a non-linear relationship between the "raw" waveform data stored by the sensor system and the actual input power. However, to establish standardized methods for the post processing of waveform data, input data related linearly to the power input are required. For some commercially available systems, this problem is handled by using a look-up table (LUT) as a transfer function from the "raw" amplitude (stored by the sensor system) of the peaks of the waveforms to their actual amplitude. Since the transformation is only valid for the peaks of the waveform, the question arises how this transformation would perturbate the shape (i.e. position, width and amplitude) of a backscattered laser pulse if applied to the whole waveform. This paper discusses the effects of the use of such non-linear transfer functions on complex laser scanner waveforms.

The multiscale analysis method is considered as an effective way to study the remotely sensed imageries of such complex landscape system. In order to overcome the shortcoming in applications of multiresolution segmentation, a four-step approach has been brought forward for homogeneous image-object detection. This is helpful to determinate the various optimal scales for diverse ground-objects in image segmentation and the potential meaningful image-objects fitting the intrinsic scale of the dominant landscape objects. Coastal landscape is complex system composed of a large number of heterogeneous components and some of them exhibit in homogenous areas. Satisfactory results were obtained in its applications in high-resolution remote sensing imageries on anthropo-directed area. The IODA result of man-made ground object is better than nature landscape because most man-made grounds objects have evident border. If the resolution of remotely sensed imagery is high enough to display object frontier then the image-objects detection results reveal perfect in MDF object candidates.

A complete data set is crucial for many applications of satellite images. Therefore, this paper tries to reconstruct the missing data sets by combining Empirical Orthogonal Functions(EOF) decomposition with Kriging methods. The EOF-based method is an effective way of reconstructing missing data for large gappiness and can maintain the macro-scale and middle-scale information of oceanographic variables. As for sparse data area (area without data or with little data all the time), EOF-based method breaks down, while Kriging interpolation turns effective. Here are the main procedures of EOF-Kriging(EOF-K) method: firstly, the data sets are processed by the EOF decomposition and the spatial EOFs and temporal EOFs are obtained; then the temporal EOFs are analyzed with Singular Spectrum Analysis(SSA); thirdly, the sparse data area is interpolated in the spatial EOFs by using Kriging interpolation; lastly, the missing data is reconstructed by using the modified spatial-temporal EOFs. Furthermore, the EOF-K method has been applied to a large data set, i.e. 151 daily Sea Surface Temperature satellite images of the East China Sea and its adjacent areas. After reconstruction with EOF-K, comparing with original data sets, the root mean square error (RMSE) of cross-validation is 0.58 °C, and comparing with in-situ Argo data, the RMSE is 0.68 °C. Thus, it has been proved that EOF-K reconstruction method is robust for reconstructing satellite missing data.

TZ-SCAN is a simple and low cost thermal imaging device which consists of a single point radiation thermometer on a tripod with a pan-tilt rotator, a DC motor controller board with a USB interface, and a laptop computer for rotator control, data acquisition, and data processing. TZ-SCAN acquires a series of zig-zag scanned data and stores the data as CSV file. A 2-D thermal distribution image can be retrieved by using the second quefrency peak calculated from TZ-SCAN data. An experiment is conducted to confirm the validity of the thermal retrieval algorithm. The experimental result shows efficient accuracy for 2-D thermal distribution image retrieval.

COnstellation of small Satellites for Mediterranean basin Observation (COSMO-SkyMed) is the largest Italian investment in Space Systems for Earth Observation, commissioned and funded by Italian Space Agency (ASI) and Italian Ministry of Defence (MoD). COSMO-SkyMed is a Dual-Use (Civilian and Defence) end-to-end Earth Observation System aimed at establishing a global service supplying provision of data, products and services compliant with well-established international standards and relevant to a wide range of applications, such as Risk Management, Scientific and Commercial Applications and Defence Applications. The system consists of a constellation of four LEO mid-sized satellites, each equipped with a multi-mode high-resolution SAR operating at X-band. The first and second COSMO-SkyMed satellites have been successfully launched the 8th of June and the 9th of December 2007 respectively, while the remaining two satellites will be gradually deployed by the end of 2009. COSMO-SkyMed 1 and 2 completed their Commissioning phase to test, verify and qualify the overall System and from the 1st of August both satellites are in the operational phase. The results of the commissioning phase of COSMO-SkyMed 1 and 2 are presented together with the Scientific Mission Exploitation strategy (i.e.: Announcement of Opportunity, Background Mission).

Image alignment is without doubt the most crucial step in SAR Interferometry. Interferogram formation requires images to be coregistered with an accuracy of better than 1/8 pixel to avoid significant loss of phase coherence. Conventional interferometric precise coregistration methods for full-resolution SAR data (Single-Look Complex imagery, or SLC) are based on the cross-correlation of the SLC data, either in the original complex form or as squared amplitudes. Offset vectors in slant range and azimuth directions are computed on a large number of windows, according to the estimated correlation peaks. Then, a two-dimensional polynomial of a certain degree is usually chosen as warp function and the polynomial parameters are estimated through LMS fit from the shifts measured on the image windows. In case of rough topography and long baselines, the polynomial approximation for the warp function becomes inaccurate, leading to local misregistrations. Moreover, these effects increase with the spatial resolution and then with the sampling frequency of the sensor, as first results on TerraSAR-X interferometry confirm. An improved, DEM-assisted image coregistration procedure can be adopted for providing higher-order prediction of the offset vectors. Instead of estimating the shifts on a limited number of patches and using a polynomial approximation for the transformation, this approach computes pixel by pixel the correspondence between master and slave by using the orbital data and a reference DEM. This study assesses the performance of this approach with respect to the standard procedure. In particular, both analytical relationships and simulations will evaluate the impact of the finite vertical accuracy of the DEM on the final coregistration precision for different radar postings and relative positions of satellites. The two approaches are compared by processing real data at different carrier frequencies and using the interferometric coherence as quality figure.

Analyses in this study show that measurements under currently used definitions on SAR image quality measurement may be unsuitable for UWB SAR. The main objective of this paper is therefore to propose a definition based on the shape of a single point target in a SAR image which is more suitable for UWB SAR. We use both real and simulated data based on the airborne UWB low frequency SAR CARABAS-II in experiments. The time-domain algorithm Global Backprojection (GBP) is selected for the image formation in this study.

Urban areas represent one of the most dynamic regions on earth. To solve the problems which are associated with the rapid changes in those regions urban and spatial planning relies on up-to-date information about the urban sprawl. Remote sensing data and in particular Synthetic Aperture Radar (SAR) images can provide valuable information about the characteristics of urban sprawl with a short repetition rate. In previous studies we could demonstrate the ability to classify built-up areas with a high accuracy using TerraSAR-X images. This paper focuses on the transfer of this method to ALOS/PALSAR images. First results of our study show that the method developed for X-band imagery can be transferred to L-band SAR images. However, the analysis of L-band data still requires some modifications of the proposed procedure in order to increase the accuracy.

Sensor and ground measurements can be considered as different representation of soil moisture characteristics. Sensor measurements, i.e. radar acquisitions, are synoptic and can cover wide areas but are influenced by other parameters such as roughness and vegetation. On the other side ground measurements are more related to soil moisture but because of the large variability of soil moisture often require some interpolation techniques to be later compared to satellite data. This paper presents an approach denominated Group Inversion Approach which aims at detecting the stable soil moisture characteristics starting from all kind of measurements available. Being reliable these points can be subsequently used to infer information about the mean soil moisture content over an area, such as a whole watershed. The soil moisture estimated from different inversion approaches are compared among them and with the ground data in relation with field characteristics. In this preliminary analysis the field characteristics have been extracted by the SAR images considering their semi-variograms. The preliminary results indicate that those fields with lowest inversion errors individuate mean field soil moisture values in good agreement with mean watershed soil moisture values.

In this paper, the Titan lake features detected by active and passive sensors during the Cassini flybys have been analyzed. The first part of the work is dedicated to the analysis of the SAR data. The main aim of this part is to find situations where it could be feasible to infer information about the likely ranges of lake parameters. The approach is mainly divided in the following steps: • Electromagnetic modeling of the lakes surface and comparison with measured radar data; • Introduction of the electromagnetic modeling in the inversion procedure based on a Bayesian framework. The inversion procedure itself is divided into training and test phase. • Analysis of the inferred information and determination of the likely ranges for lake optical thickness and wind speed. In the second part, the analysis have been extended to the radiometer data. The likely ranges of surface parameters e.g. dielectric constant, wind speed and optical thickness derived from the Bayesian inversion have been used as an input to a forward radiative transfer model calculation to obtain simulated brightness temperatures. Comparison of the observed and computed brightness temperatures will allow us to address the consistency of the observations from the two instruments and to determine the coarse characteristics of the surface parameters that lead to consistent results between the two instruments at a global scale. In processing radiometric data two main aspects have to be considered. The correlation between SAR and radiometry is limited by the low radiometry footprint resolution at closest approach, preventing detection and correlation of surface features smaller than 6 km azimuth resolution. In addition, there is a large-scale variability in calibration among the five radiometer beams caused by sidelobes that should be accounted for obtaining relative brightness variations. For these two reasons, the data have been compared only on areas with large extensions such as large liquid surfaces detected during T25 and T30 Cassini flybys and relative surrounding areas. Furthermore only relative brightness temperature, calculated as difference between various areas, have been considered.

Twenty five ASAR scenes are analyzed to find the suitable pairs for generating DEM of Kuwait desert area. About 40 pairs are suitable for generating DEMs. A unique combination of seven DEM are possible with a single master image whose perpendicular baseline component varies between 233 to 393m. GAMMA inerferometric package coupled with ERDAS image processing package are used in the analysis. The seven DEMs are compared with the 90 m DEM derived from SRTM. It has been found that the RMS error varies from 1.9 m to 15.3 m. The highest RMS error refers to day-difference of 525 days with average coherence value of 0.71 (lowest of all). Therefore, correlation will be the cause of higher errors. However, 35 days day-difference pair gave RMS error of 5.5 m with highest coherence value (0.93) which is supposed to produce the lowest RMS error. For the study of spatial distribution of errors, Interferometric DEMs are subtracted from DEM of SRTM. From this analysis, it is observed that the atmospheric affects are aligned and varies systematically. It can give an error in elevation as high as 15 m. Therefore, one should be very careful in using SAR interfoermetry technology for DEM generation even for desert regions.

Nowadays, several free or commercial tools for INSAR, DINSAR and even PS INSAR exist. A brief inventory of current suites is drawn and shows their main capabilities. The objective of this work is the handling and the validation of our use of DORIS, a free processing suite. For this purpose, the organization of the processing is studied and compared to another one (SARSCAPE, a commercial software). It appears that both chains contain identical or similar steps but, some stages are missing or are replaced by a different one. Experiments are then held with DORIS and SARSCAPE using the same data set: two ASAR images covering the 2003 Bam (Iran) earthquake and a SRTM3 DEM. Using DORIS and SARSCAPE, it leads to comparable results, which allows validating our use of DORIS (and also SARSCAPE). However, linear and non linear residual differences occur. In the same way, dissimilarities are observed with other published results. These dispecrancies are interpreted as resulting from residual geometric inaccuracies. One solution to reduce these artefacts is a simple surface compensation but the orbital parameters are then not corrected. Finally, this work may give some help for a better understanding of any DINSAR processing chain.

SAR simulation tool is useful for the design and implementation of SAR systems. The effects of varying system parameters may be easily investigated through simulation. Raw signal generation of SAR embodies the distributed surface model, layover, shadow and scattering mechanism. We use the facet model to have the normal vector toward the space from earth. It gives us incident angles of the objects, which provides reflectivity of whole scene. We also detect layover, shadow area by using the nominal collection look/azimuth angle form, which is derived from the distributed surface model.

In this paper, we propose a new ship detection model on SAR imagery. The model uses the Pearson distribution system to simulate the backscattering distribution from ocean surface on SAR imagery. In the Pearson distribution system, four distributions including the pearson distributions of type I, III, IV and VI are employed in the model. Using these four distributions, we form a CFAR equation. A distribution selected machine basedβplane is used to select which distribution is adopted for specify SAR imagery. We can get four equations and get the threshold of gray level. Then, using the threshold, the model can find ships from SAR images.