Image thresholding based on the EM algorithm and the generalized Gaussian distribution

In this paper, a novel parametric and global image histogram thresholding method is presented. It is based on the estimation of the statistical parameters of “object” and “background” classes by the expectation-maximization (EM) algorithm, under the assumption that these two classes follow a generalized Gaussian (GG) distribution. The adoption of such a statistical model as an alternative to the more common Gaussian model is motivated by its attractive capability to approximate a broad variety of statistical behaviors with a small number of parameters. Since the quality of the solution provided by the iterative EM algorithm is strongly affected by initial conditions (which, if inappropriately set, may lead to unreliable estimation), a robust initialization strategy based on genetic algorithms (GAs) is proposed. Experimental results obtained on simulated and real images confirm the effectiveness of the proposed method.     

Classification of hyperspectral remote sensing images with support vector machines

This paper addresses the problem of the classification of hyperspectral remote sensing images by support vector machines (SVMs). First, we propose a theoretical discussion and experimental analysis aimed at understanding and assessing the potentialities of SVM classifiers in hyperdimensional feature spaces. Then, we assess the effectiveness of SVMs with respect to conventional feature-reduction-based approaches and their performances in hypersubspaces of various dimensionalities. To sustain such an analysis, the performances of SVMs are compared with those of two other nonparametric classifiers (i.e., radial basis function neural networks and the K-nearest neighbor classifier). Finally, we study the potentially critical issue of applying binary SVMs to multiclass problems in hyperspectral data. In particular, four different multiclass strategies are analyzed and compared: the one-against-all, the one-against-one, and two hierarchical tree-based strategies. Different performance indicators have been used to support our experimental studies in a detailed and accurate way, i.e., the classification accuracy, the computational time, the stability to parameter setting, and the complexity of the multiclass architecture. The results obtained on a real Airborne Visible/Infrared Imaging Spectroradiometer hyperspectral dataset allow to conclude that, whatever the multiclass strategy adopted, SVMs are a valid and effective alternative to conventional pattern recognition approaches (feature-reduction procedures combined with a classification method) for the classification of hyperspectral remote sensing data.     

Detection of premature ventricular contraction arrhythmias in electrocardiogram signals with kernel methods

In this paper, we propose to investigate the capabilities of two kernel methods for the detection and classification of premature ventricular contractions (PVC) arrhythmias in Electrocardiogram (ECG signals). These kernel methods are the support vector machine and Gaussian process (GP). We propose to study these two classifiers with various feature representations of ECG signals, such as morphology, discrete wavelet transform, higher-order statistics, and S transform. The experimental results obtained on 48 records (i.e., 109,887 beats) of the MIT-BIH Arrhythmia database showed that for all feature representation adopted in this work, the GP detector trained only with 600 beats from PVC and Non-PVC classes can provide an overall accuracy and a sensitivity above 90 % on 20 records (i.e., 49,774 beats) and 28 records (i.e., 60,113 beats) seen and unseen, respectively, during the training phase.     

A two-stage regression approach for spectroscopic quantitative analysis

In this paper, we propose a two-stage regression approach, which is based on the residual correction concept. Its underlying idea is to correct any given regressor by analyzing and modeling its residual errors in the input space. We report and discuss results of experiments conducted on three different datasets in infrared spectroscopy and designed in such a way to test the proposed approach by: 1) varying the kind of adopted regression method used to approximate the chemical parameter of interest. Partial least squares regression (PLSR), support vector machines (SVM) and radial basis function neural network (RBF) methods are considered; 2) adopting or not a feature selection strategy to reduce the dimension of the space where to perform the regression task. A comparative study with another approach which exploits differently estimation errors, namely adaptive boosting for regression (AdaBoost.R), is also included. The obtained results point out that the residual-based correction approach (RBC) can improve the accuracy of the estimation process. Not all the improvements are statistically significant but, at the same time, no case of accuracy decrease has been observed.     

New fusion and selection approaches for estimating the remaining useful life using Gaussian process regression and induced ordered weighted averaging operators

In this paper, we propose new fusion and selection approaches to accurately predict the remaining useful life. The fusion scheme is built upon the combination of outcomes delivered by an ensemble of Gaussian process regression models. Each regressor is characterized by its own covariance function and initial hyperparameters. In this context, we adopt the induced ordered weighted averaging as a fusion tool to achieve such combination. Two additional fusion techniques based on the simple averaging and the ordered weighted averaging operators besides a selection approach are implemented. The differences between adjacent elements of the raw data are used for training instead of the original values. Experimental results conducted on lithium-ion battery data report a significant improvement in the obtained results. This work may provide some insights regarding the development of efficient intelligent fusion alternatives for further prognostic advances.     

Using convolutional features and a sparse autoencoder for land-use scene classification

In this article, we propose a novel approach based on convolutional features and sparse autoencoder (AE) for scene-level land-use (LU) classification. This approach starts by generating an initial feature representation of the scenes under analysis from a deep convolutional neural network (CNN) pre-learned on a large amount of labelled data from an auxiliary domain. Then these convolutional features are fed as input to a sparse AE for learning a new suitable representation in an unsupervised manner. After this pre-training phase, we propose two different scenarios for building the classification system. In the first scenario, we add a softmax layer on the top of the AE encoding layer and then fine-tune the resulting network in a supervised manner using the target training images available at hand. Then we classify the test images based on the posterior probabilities provided by the softmax layer. In the second scenario, we view the classification problem from a reconstruction perspective. To this end we train several class-specific AEs (i.e. one AE per class) and then classify the test images based on the reconstruction error. Experimental results conducted on the University of California (UC) Merced and Banja-Luka LU public data sets confirm the superiority of the proposed approach compared to state-of-the-art methods.