Charalampos Topaloglou

Decision Fusion of GA Self-Organizing Neuro-Fuzzy Multilayered Classifiers for Land Cover Classification Using Textural and Spectral Features

A novel Self-Organizing Neuro-Fuzzy Multilayered Classifier, the GA-SONeFMUC model, is proposed in this paper for land cover classification of multispectral images. The model is composed of generic fuzzy neuron classifiers (FNCs) arranged in layers, which are implemented by fuzzy rule-based systems. At each layer, parent FNCs are combined to generate a descendant FNC at the next layer with higher classification accuracy. To exploit the information acquired by the parent FNCs, their decision supports are combined using a fusion operator. As a result, a data splitting is devised within each FNC, distinguishing those pixels that are currently correctly classified to a high certainty grade from the ambiguous ones. The former are handled by the fuser, while the ambiguous pixels are further processed to enhance their classification confidence. The GA-SONeFMUC structure is determined in a self-constructing way via a structure-learning algorithm with feature selection capabilities. The parameters of the models obtained after structure learning are optimized using a real-coded genetic algorithm. For effective classification, we formulated three input sets containing spectral and textural feature types. To explore information coming from different feature sources, we apply a classifier fusion approach at the final stage. The outputs of individual classifiers constructed from each input set are combined to provide the final assignments. Our approach is tested on a lake-wetland ecosystem of international importance using an IKONOS image. A high-classification performance of 92.02% and of 75.55% for the wetland zone and the surrounding agricultural zone is achieved, respectively.

A novel self-organizing neuro-fuzzy multilayered classifier for land cover classification of a VHR image

A novel self‐organizing neuro‐fuzzy multilayered classifier (SONeFMUC) is introduced in this paper, with feature selection capabilities, for the classification of an IKONOS image. The structure of the proposed network is developed in a sequential fashion using the group method of data handling (GMDH) algorithm. The node models, regarded as generic classifiers, are represented by fuzzy rule‐based systems, combined with a fusion scheme. A data splitting mechanism is incorporated to discriminate between correctly classified and ambiguous pixels. The classifier was tested on the wetland of international importance of Lake Koronia, Greece, and the surrounding agricultural area. To achieve higher classification accuracy, the image was decomposed into two zones: the wetland and the agricultural zones. Apart from the initial bands, additional input features were considered: textural features, intensity–hue–saturation (IHS) and tasseled cap transformation. To assess the quality of the suggested model, the SONeFMUC was compared with a maximum likelihood classifier (MLC). The experimental results show that the SONeFMUC exhibited superior performance to the MLC, providing less confusion of the dominant classes in both zones. In the wetland zone, an overall accuracy of 89.5% was attained.

A neuro-fuzzy multilayered classifier for land cover image classification

In this paper, a novel Self-Organizing Neuro-Fuzzy Multilayered Classifier (SONeFMUC) is suggested for land cover classification of a SPOT-5 satellite image. The proposed model is developed in a self-organizing manner by means of the group method of data handling (GMDH) algorithm, exhibiting feature selection capabilities. Each node is regarded as a generic fuzzy neuron classifier (FNC) which is implemented by fuzzy rule-based systems, combined with a decision fusion scheme. A data splitting mechanism is incorporated to discriminate between confident classified and ambiguous pixels, providing an efficient handling of the data flow. The application of the model was performed to the agricultural area of Larisa, Greece. Apart from the initial bands, additional features were used, namely intensity, hue and saturation transformation. The classification performance and the thematic map produced by SONeFMUC demonstrate the classification capabilities of the proposed model.

Monitoring water quality parameters in the marine area of Danube Delta using satellite remote sensing: preliminary results

The aim of this work was to produce water quality parameter maps for the marine area of the Danube Delta using remotely sensed data and to validate the results with in-situ measurements. For this reason, satellite images from ENVISAT/MERIS and Aqua/MODIS were used along with collocated in-situ measurements. The latter were in-sync with the satellite images acquisition so that rigorous and validation could be performed. Chlorophyll-a concentration and total suspended matter were estimated using the CASE-II algorithm and MERIS satellite images, while sea surface temperature was estimated from MODIS Ocean Team products. The results show that the satellite images covered the study area completely, with some data gaps due to cloud coverage. Comparisons show a good correspondence with in-situ measurements. Thus, the time series of satellite images that was produced suggests that it is possible to monitor the biological changes on an operational basis. The produced maps described a detailed spatial pattern of chlorophyll-a and total suspended matter that could not have been identified from the sparse in-situ measurements.

Modeling LAI based on land cover map and NDVI using SPOT and Landsat data in two Mediterranean sites: preliminary results

Leaf Area Index (LAI) is considered to be a key parameter of ecosystem processes and it is widely used as input to biogeochemical process models that predict net primary production (NPP) or can be a useful parameter for crop yield prediction and crop stress assessment as well as estimation of the exchanges of carbon dioxide, water, and nutrients in forests. LAI can be derived from satellite optical data using models referred to physical-based approaches, which describe the physical processes of energy flow in the soil-vegetation-atmosphere system, and models using empirically derived regression relationships based on spectral vegetation indices (VIs). The first category of models are more general in application because they can account for the different sources of variability, although in many cases the information needed to constrain model inputs is not available. In contrast, empirical models depend on the site and time. The aim of this paper is to create a reliable semi-empirical method, applied in two Mediterranean sites, to estimate LAI with high spatial resolution images. The model uses a minimum dataset of a Landsat 5 TM or SPOT 4 XS image, land cover map and DEM for each area. Specifically, this model calculates the reflectance of initial bands implementing topographic correction with the aid of DEM and metadata of the images and afterwards uses a list of NDVI values that correspond to certain LAI values on different land cover types which has been proposed by the MODIS Land Team. This model has been applied in two areas; in the river basin of Nestos (Greece and Bulgaria) and in the river basin of Tamega (Portugal). The predicted LAI map was validated with ground truth data from hemispherical images showing high correlation, with r reaching 0.79 and RMSE less than 1 m2/m2.

Accurate crop classification using hierarchical genetic fuzzy rule-based systems

This paper investigates the effectiveness of an advanced classification system for accurate crop classification using very high resolution (VHR) satellite imagery. Specifically, a recently proposed genetic fuzzy rule-based classification system (GFRBCS) is employed, namely, the Hierarchical Rule-based Linguistic Classifier (HiRLiC). HiRLiC’s model comprises a small set of simple IF–THEN fuzzy rules, easily interpretable by humans. One of its most important attributes is that its learning algorithm requires minimum user interaction, since the most important learning parameters affecting the classification accuracy are determined by the learning algorithm automatically. HiRLiC is applied in a challenging crop classification task, using a SPOT5 satellite image over an intensively cultivated area in a lake-wetland ecosystem in northern Greece. A rich set of higher-order spectral and textural features is derived from the initial bands of the (pan-sharpened) image, resulting in an input space comprising 119 features. The experimental analysis proves that HiRLiC compares favorably to other interpretable classifiers of the literature, both in terms of structural complexity and classification accuracy. Its testing accuracy was very close to that obtained by complex state-of-the-art classification systems, such as the support vector machines (SVM) and random forest (RF) classifiers. Nevertheless, visual inspection of the derived classification maps shows that HiRLiC is characterized by higher generalization properties, providing more homogeneous classifications that the competitors. Moreover, the runtime requirements for producing the thematic map was orders of magnitude lower than the respective for the competitors.