Vassilia Karathanassi

Investigation of the Dual-Tree Complex and Shift-Invariant Discrete Wavelet Transforms on Quickbird Image Fusion

In the current survey, the performance of the shift-invariant discrete wavelet transform and dual-tree complex wavelet transform (DT-CWT) for Quickbird image fusion is investigated. For this purpose, a DT-CWT fusion algorithm is developed and implemented on high-resolution multispectral and panchromatic Quickbird images of Heraclion, Crete, Greece. In order to point out the effectiveness of the aforementioned transforms, the resulting imagery is visually (through photointerpretation) and computationally (through index computations) compared to fusion products derived by other commonly used methods, such as the intensity hue saturation transform (IHS), the discrete wavelet transform, and the crossbred wavelet and IHS transform. The DT-CWT has been proved to provide a complete and effective tool for Quickbird image fusion

Potentiality of feed-forward neural networks for classifying dark formations to oil spills and look-alikes

Radar backscatter values from oil spills are very similar to backscatter values from very calm sea areas and other ocean phenomena. Several studies aiming at oil spill detection have been conducted. Most of these studies rely on the detection of dark areas, which have high Bayesian probability of being oil spills. The drawback of these methods is a complex process, mainly because non-linearly separable datasets are introduced in statistically based decisions. The use of neural networks (NNs) in remote sensing has increased significantly, as NNs can simultaneously handle non-linear data of a multidimensional input space. In this article, we investigate the ability of two commonly used feed-forward NN models: multilayer perceptron (MLP) and radial basis function (RBF) networks, to classify dark formations in oil spills and look-alike phenomena. The appropriate training algorithm, type and architecture of the optimum network are subjects of research. Inputs to the networks are the original synthetic aperture radar image and other images derived from it. MLP networks are recognized as more suitable for oil spill detection.

An Unsupervised Classification Approach for Polarimetric SAR Data Based on the Chernoff Distance for Complex Wishart Distribution

A new unsupervised classification approach for polarimetric synthetic aperture radar (POLSAR) data is proposed in this paper. The Wishart-Chernoff distance is calculated and used in an agglomerative hierarchical clustering approach. Initial segmentation of POLSAR data into clusters is obtained based on the total backscattering power (SPAN) combined with the entropy, alpha angle, and anisotropy. The complex Wishart clustering is performed to optimize the initialization. Optimized clusters with minimum Wishart-Chernoff distance are merged hierarchically into an appropriate number of classes. The appropriate number of classes is estimated based on the data log-likelihood algorithm. Classification results show that the use of Wishart-Chernoff distance is superior to that of the Wishart test statistic distance. The effectiveness of the proposed Wishart-Chernoff distance is demonstrated using Advanced Land Observing Satellite POLSAR data.

Large-scale feature selection using evolved neural networks

In this paper computational intelligence, referring here to the synergy of neural networks and genetic algorithms, is deployed in order to determine a near-optimal neural network for the classification of dark formations in oil spills and look-alikes. Optimality is sought in the framework of a multi-objective problem, i.e. the minimization of input features used and, at the same time, the maximization of overall testing classification accuracy. The proposed method consists of two concurrent actions. The first is the identification of the subset of features that results in the highest classification accuracy on the testing data set i.e. feature selection. The second parallel process is the search for the neural network topology, in terms of number of nodes in the hidden layer, which is able to yield optimal results with respect to the selected subset of features. The results show that the proposed method, i.e. concurrently evolving features and neural network topology, yields superior classification accuracy compared to sequential floating forward selection as well as to using all features together. The accuracy matrix is deployed to show the generalization capacity of the discovered neural network topology on the evolved sub-set of features.

Estimation of the Number of Endmembers Using Robust Outlier Detection Method

This paper introduces a novel approach for estimating the numbers of endmembers in hyperspectral imagery. It exploits the geometrical properties of the noise hypersphere and considers the signal as outlier of the noise hypersphere. The proposed method, called outlier detection method (ODM), is automatic and non-parametric. In a principal component space, noise is spherically symmetric in all directions and lies on the surface of a hypersphere with a constant radius. Reversely, signal radiuses are much larger that noise radius and vary in all directions, thus signal lies in a hyperellipsoid. The proposed method involves three steps: 1) noise estimation; 2) minimum noise fraction transformation; and 3) outlier detection using inter quartile range. Estimation of the number of endmembers is accomplished by the estimation of the number of noise hypersphere outliers using a robust outlier detection method. The ODM was evaluated using simulated and real hyperspectral data, and it was also compared with well-known methods for estimating the number of endmembers. Evaluation of the method showed that the method produces robust and satisfactory results, and outperforms in relation to its competitors.

A knowledge-based classification method for polarimetric SAR data

Full polarimetric data can define the scattering behavior of land use/cover through several approaches. Several classification methods have been proposed based on analysis methods. These classification methods are based on the backscattering mechanisms which are extracted using a single decomposition method. The objectives of this work are a) the investigation of the different polarimetric analysis methods; b) the interpretation of the images resulting from polarimetric analysis; c) the development of an object-oriented classification method based on polarimetric analysis imagery and the comparison of this method with the H/a and the H/a Wishart classification methods respectively.

A multi-level segmentation methodology for dual-polarized SAR data

Abstract An innovative methodology for dual-polarized Synthetic Aperture Radar (SAR) data segmentation is proposed. The methodology is based on the thresholding of the 1D-histograms of the two images produced by the dual polarimetric bands. Thresholding of the histograms is performed using a nonparametric algorithm. Histograms after thresholding are combined together in a two dimensional histogram-based space in order to define sub-spaces, which are used for image segmentation. Sub-spaces are further divided based on two criteria which lead to a multi-level segmentation approach. Dual-polarized TerraSAR-X data, both HH and VV, are used in a study area located in the southwestern United Kingdom.

Vegetation biomass estimation with remote sensing: focus on forest and other wooded land over the Mediterranean ecosystem

ABSTRACT Carbon sequestration service of Mediterranean forest and other wooded land is threatened by their fragile, complex, and highly evolving nature, due to both human disturbances and climate change. Remote-sensing methods for forest biomass estimation have gained increased attention, and substantial research has been conducted worldwide over the past four decades. Yet, the literature body focused on Mediterranean forests is rather limited as a result of their small extent compared to other biomes. We discuss the remote-sensing studies over the Mediterranean forest and other wooded land, discriminating research based on the primary data source used, such as optical imagery, datasets from active sensors, and combination of multisource data. The review indicates that there is a significant research gap in terms of the studies, as well as a need for a reduction of the errors and uncertainty of estimates, which are associated with both the sensors’ characteristics and the Mediterranean forest and other wooded land structure. Biomass estimates based on optical data were generally less accurate (R2 close to 0.70, where R2 is the coefficient of determination), however, when data from active sensors were involved, accuracy of estimations was considerably greater (usually R2 greater than 0.80). With respect to scale, most of the local scale studies established relationships with R2 over 0.70 and as high as 0.98, while the few regional scale studies exhibited R2 close to 0.80. Further, in-depth analysis can provide more efficient data fusion, classification methods, and procedures for operational regional and national assessment of forest biomass over such Mediterranean areas.

Development of a Network-Based Method for Unmixing of Hyperspectral Data

This paper presents a new nonlinear unmixing method. Based on relative distances which imply nonlinearity, the method introduces the “fractional distance” as a key variable that quantifies interactions between pixels and endmembers. Relationships between fractional distances and abundance fractions are built through networks. Because an equal spectral mixture of ground spectral classes present on the surface sensed is likely impossible, the proposed method, due to its mathematical concept, reveals unknown endmembers. Three versions of the method have been developed: the nonconstrained, the sum-to-one, and the fully constrained versions. Evaluation of the method using synthetic and real data showed that the method is robust with clear and interpretable results and provides reliable abundance fractions, particularly the sum-to-one and the fully constrained versions of the method. The new unmixing method has also been compared with the fully constrained least squares method.

Investigation of hyperspectral remote sensing for mapping asphalt road conditions

An investigation of hyperspectral remote sensing for mapping asphalt road conditions is undertaken in this study. Hyperspectral data acquired by the GER1500 radiometer and the Compact Airborne Spectrographic Imager (CASI) 550 sensor have been analysed, processed and interpreted. Field radiometer data were used to provide high-quality spectral measurements for developing a spectral library for asphalt, defining potential categories of the asphalt condition and minimizing the dimension of the hyperspectral space. Analysis of spectral signatures indicated that asphalt condition is affected by asphalt age, material quality and road circulation, and that it led to the definition of five potential categories. Two of them indicate asphalt in high distress and surfaces that need rehabilitation. Among several others, the following processing methods were revealed as the most suitable for detecting asphalt condition: Principal Component Analysis (PCA), thresholding of colour transformation images, unsupervised classification Iterative Self-organizing Data Analysis (IsoData), supervised classification Spectral Angle Mapper (SAM) and texture measurements using the Grey-level Co-occurrence Matrix operator. The results indicated that hyperspectral remote sensing is capable of mapping asphalt road conditions with respect to the categorization proposed within this study.