Chiara Pratola

Toward Fully Automatic Detection of Changes in Suburban Areas From VHR SAR Images by Combining Multiple Neural-Network Models

Recent X-band SAR missions, such as COSMO-SkyMed (CSK), which is able to provide very high spatial resolution images of an area of interest with a short revisit time, are expected to be quite useful sources of information for monitoring the terrestrial environment and its changes. On the other hand, the huge amount of data involved, as well as the need to promptly act in case of emergency, requires the development of automatic change detection tools. This paper reports on a novel automatic change detection algorithm combining multilayer perceptron neural networks (NNs) and pulse coupled NNs, which has been implemented and tested on pairs of Stripmap and Spotlight CSK images acquired on the Tor Vergata University area in the southeast outskirts of Rome, Italy, where a significant and continuous urbanization process is occurring.

Evaluation of a global soil moisture product from finer spatial resolution sar data and ground measurements at Irish sites

In the framework of the European Space Agency Climate Change Initiative, a global, almost daily, soil moisture (SM) product is being developed from passive and active satellite microwave sensors, at a coarse spatial resolution. This study contributes to its validation by using finer spatial resolution ASAR Wide Swath and in situ soil moisture data taken over three sites in Ireland, from 2007 to 2009. This is the first time a comparison has been carried out between three sets of independent observations from different sensors at very different spatial resolutions for such a long time series. Furthermore, the SM spatial distribution has been investigated at the ASAR scale within each Essential Climate Variable (ECV) pixel, without adopting any particular model or using a densely distributed network of in situ stations. This approach facilitated an understanding of the extent to which geophysical factors, such as soil texture, terrain composition and altitude, affect the retrieved ECV SM product values in temperate grasslands. Temporal and spatial variability analysis provided high levels of correlation (p < 0.025) and low errors between the three datasets, leading to confidence in the new ECV SM global product, despite limitations in its ability to track the driest and wettest conditions.

PCNN for automatic segmentation and information extraction from X-band SAR imagery

The extremely high number of synthetic aperture radar (SAR) images provided by the current spaceborne missions demand for the development of even more effective automatic techniques for data processing. In this context, neural approaches can give significant contributions being characterised by a high level of automatism. In particular, rather interesting potential is provided by the pulse-coupled neural networks (PCNNs), which have been designed with the idea of simulating the visual cortex of small mammals. In this article, the performance of PCNNs for automatic object extraction from satellite with very high-resolution SAR images is examined by applying them to different cases of interest.

TerraSAR-X/SPOT-5 Fused Images for Supervised Land Cover Classification

This paper reports the study of supervised neural network algorithm for classification purposes. SPOT 5 and TerraSAR-X dataset are analyzed. Classification results are critically discussed and compared to ground truth map and unsupervised neural classification of the same area. The aim is to demonstrate the capability of neural networks in managing heterogeneous dataset and the accuracy improvement obtained by the use of the textural object based layers fused with the optical and radar data.

Quality Assessment of the CCI ECV Soil Moisture Product Using ENVISAT ASAR Wide Swath Data over Spain, Ireland and Finland

During the last decade, great progress has been made by the scientific community in generating satellite-derived global surface soil moisture products, as a valuable source of information to be used in a variety of applications, such as hydrology, meteorology and climatic modeling. Through the European Space Agency Climate Change Initiative (ESA CCI), the most complete and consistent global soil moisture (SM) data record based on active and passive microwaves sensors is being developed. However, the coarse spatial resolution characterizing such data may be not sufficient to accurately represent the moisture conditions. The objective of this work is to assess the quality of the CCI Essential Climate Variable (ECV) SM product by using finer spatial resolution Advanced Synthetic Aperture Radar (ASAR) Wide Swath and in situ soil moisture data taken over three regions in Europe. Ireland, Spain, and Finland have been selected with the aim of assessing the spatial and temporal representativeness of the ECV SM product over areas that differ in climate, topography, land cover and soil type. This approach facilitated an understanding of the extent to which geophysical factors, such as soil texture, terrain composition and altitude, affect the retrieved ECV SM product values. A good temporal and spatial agreement has been observed between the three soil moisture datasets for the Irish and Spanish sites, while poorer results have been found at the Finnish sites. Overall, the two different satellite derived products capture the soil moisture temporal variations well and are in good agreement with each other.

Characterizing land cover from X-band COSMO-SkyMed images by neural networks

The launch of last-generation satellites (COSMO-SkyMed and TerraSAR-X), equipped with X-band sensors acquiring images with a very high spatial resolution, has opened up new challenges in the field of SAR image processing for remote sensing applications. In this work, a set of Spotlight and Stripmap COSMO-Skymed images taken the Tor Vergata-Frascati test site was considered to investigate on the potential of this type of data in characterizing sub-urban areas by exploiting both amplitude and phase information contained in the radar return. In particular, this contribution deals with the development of a pixel based classification technique based on Multi-Layer Perceptron (MLP) Neural Networks (NN). The results have been compared with a land cover map of the same area, achieved by means of a different neural network algorithm exploiting the information carried by the eight bands of WorldView-2 satellite.

Automatic features extraction in sub-urban landscape using very high resolution Cosmo-Skymed SAR images

The new generation of spaceborne instruments, capable of capturing a large amount of very-high resolution images within a short revisit time, is allowing remote sensing researchers and final users to receive huge amounts of data in rather short times. Such a scenario makes it mandatory the development of techniques, as much as possible automatic, for the understanding and the effective exploitation of the available information. This contribution deals with the features extraction from Spotlight Cosmo-SkyMed SAR imagery (1 m spatial resolution) by means Multi Layer Perceptron Neural Network (MLP-NN) algorithms. For a better pixel characterization, textural parameters have been also considered as additional information for the classification procedure.

Classification of PingPong COSMO-SkyMed imagery using supervised and unsupervised neural network algorithms

The novel instruments of the COSMO-SkyMed (CSK) Earth Observation programme, offer an opportunity to explore at various resolutions the information content of X-band signal backscattered with different polarizations. In spite of their potential to render additional information about an area of interest, speckle noise and artifacts make X-band acquisitions difficult to interpret. This is a motivating scenario to explore what (semi-)automatic procedures might be able to offer. This paper is first attempt to process CSK Stripmap PingPong data using two well-known artificial neural network techniques: the supervised backpropagation multilayer perceptron and the unsupervised self-organizing map.

Towards fully automatic generation of land cover maps from polarimetric and metric-resolution SAR data

Information mining from heavy SAR images is considered from the point of view of the procedure automatization. Two schemes based on Neural Networks are evaluated, one based on the Self Organizing Map method exploiting polarimetric information and oriented to land cover classification, the other based on the Pulse-Coupled Neural Networks aiming at characterizing the imaged buildings.

Automatic object extraction from VHR satellite SAR images using pulse coupled neural networks

In this paper we investigate an unsupervised neural network approach for automatically extracting objects of interest from very high resolution (VHR) SAR images. The technique is based on the use of Pulse-Coupled Neural Networks (PCNN) which is a relatively novel technique based on models of the visual cortex of small mammals. The study discusses the use of PCNN technique in different applications. In a first case the extraction procedure is focused on the detection of buildings. In the second case the segmentation of a dark spot representing an oil spill in a SAR image is considered. The performance yielded by the PCNN is evaluated and critically discussed for a set of new generation of X-band SAR images taken by COSMO-Skymed and TerraSAR-X systems.