A recurrent concern in cloud detection approaches is the high misclassification rate for pixels close to cloud edges. We tackle this problem by introducing a novel penalty term within the classical maximum a posteriori probability-Markov random field (MAP-MRF) approach. To improve the classification rate, such term, for which we suggest two different functional forms, accounts for the predictable motion of cloud volumes across images. Two mass tracking techniques are proposed. The first one is an effective and efficient implementation of the probability hypothesis density (PHD) filter, which is based on Gaussian mixtures (GMs) and relies on finite set statistics (FISST). The second one is a region matching procedure based on a maximum cross-correlation (MCC) that is characterized by low computational load. Through extensive tests on simulated images and real data, acquired by the SEVIRI sensor, both methods show a clear performance gain in comparison with classical spatial MRF-based algorithms.

Support Vector Machines (SVM) emerge among the classification methods as a very effective tool for separating the illuminated scene in different classes by utilizing multiple features. Typically, a pixelwise classification is performed by employing, as features, the radiances at different bands. This neglects the possibility of accounting for the spatial, and possibly the temporal, correlation within and among the images. Existing methods include the latter through segmentation methods whose output is fused with the SVM classification. We propose here to adjoin the contextual information as a further feature by constructing a penalty map accounting for the correlation among pixels. To illustrate and evaluate the method we present an application of cloud masking on MultiSpectral Images acquired by the SEVIRI sensor.

The design of multitarget tracking procedures includes, as the most time consuming steps, the definition of the objective class and the formulation of the detection criteria. In this paper we investigate a solution toward an intuitive way for implementing a detector for any ad-hoc application. We capitalize on the OSPA metric to discriminate between the semantic object class of interest and other look-alike classes starting from a short number of unlabeled markers. We propose an illustrative algorithm with a toy example, then we apply it to two real images, the first acquired by SEVIRI, the second by MERIS. In the first case we discriminate between lakes, sea and look-alike clouds, in the other between ground and sea ice. We show how semantic classes with very similar spectral properties can be separated even in the presence of uncertainties or errors in the ground truth.

Pansharpening algorithms aim to enhance low resolution multi-spectral images by means of high resolution panchromatic ones. Several approaches are based on the MultiResolution Analysis (MRA) achieved through the pyramidal decomposition of images. We focus here on the implementation based on Morphological Filters (MF) that are optimized through Genetic Algorithms (GA). The effectiveness of this algorithm is compared with other techniques, among which those based on Wavelet operators, through several quality indices on two different real scenarios.

In irrigation management the estimation of the radiometric surface temperature is of fundamental importance in evaluating the spatial distribution of land surface evapotranspiration. However, obtaining both high spatial and temporal resolutions data is impossible for any real sensor. In this paper we propose and investigate the use of sequential Bayesian techniques for integrating heterogeneous data with complementary features. A validation is performed by means of images acquired from SEVIRI and MODIS sensors in the thermal channels IR 10:8 and 31, respectively.

Temporal correlation has been recently taken into consideration to improve the performances of cloud detection algorithms. We exploit this concept within the Maximum A Posteriori Markov Random Field (MAP-MRF) framework by adding a penalty term which is determined according to the history of cloud masses. Multi Target Tracking of clouds is accomplished by methods of FInite Set STatistics (FISST) and several particle-based implementations are compared among them and with other previous methods both on simulated and real data.

In this paper we present a cloud detection algorithm exploiting both the spatial and the temporal correlation of cloudy images. A region matching technique for cloud motion estimation is embodied into a MAP-MRF framework through a penalty term. We test our proposal both on simulated data and on real images acquired by MSG satellite sensors (SEVIRI) in the VIS 0.8 band. Comparisons with classical MRF based algorithms show our approach to achieve better results in terms of misclassification probability and, in particular, to be very effective in detecting cloud edges.

Information extraction from remotely sensed images acquired in the visible and near-infrared (VNIR) frequency range strongly depends on an accurate cloud pixel screening. Indeed, many remote sensing applications require a preliminary cloud detection phase to obtain profitable results. In this paper we propose to integrate the potential of the MAP-MRF methodology with the multispectral approach for augmenting the capability of the algorithm to detect cloudy pixels. In particular the proposed technique combines information from some SEVIRI sensor channels (in particular the channels 0.64 mu m, 1.6 mu m, 3.9 mu m, 7.3 mu m and 10.8 mu m) with the classification obtained by the MAP-MRF method in the 0.8 mu m channel in order to discriminate between snowy and cloudy pixels.The validation is performed on challenging images of Alps mountains acquired by the SEVIRI sensor during winter months. Results show significant improvements with respect to existing methods. In particular we highlight a more precise classification at the cloud borders and a considerable reduction of unsolicited holes inside the cloud masses.