Classification Of Hyperspectral Image Data Research Articles

Band selection technique based on binary modified equilibrium optimizer for hyperspectral image classification

The performance of hyperspectral image (HSI) classification can be enhanced by utilizing a metaheuristic technique with high exploration and exploitation capabilities for band selection. We design a framework for the improved classification of HSIs by utilizing a binary modified equilibrium optimizer (BMEO) for band selection. First, the proposed framework preprocesses the HSI data using the Gaussian filter to remove the undesired variations. Band selection is performed on the preprocessed HSI data using a BMEO, which exhibits balanced and high exploration and exploitation capabilities. The proposed V-shape transfer function is used for enhancing the exploration capabilities of the BMEO. An analysis for band selection is performed by comparing the reflectance spectra, correlation of each band with ground truth, and correlation between the bands before and after the band selection. The classification of HSI data by utilizing selected bands was done using the classifier selected by classifier analysis. The proposed framework outperforms 11 state-of-the-art techniques, including filter, wrapper, and deep learning-based techniques. The 95.85%, 98.13%, and 98.31% overall accuracies on the Indian pine, Pavia University, and Salinas datasets, respectively, show the significance of the proposed framework.

Computational Intelligence for Observation and Monitoring: A Case Study of Imbalanced Hyperspectral Image Data Classification.

Imbalance in hyperspectral images creates a crisis in its analysis and classification operation. Resampling techniques are utilized to minimize the data imbalance. Although only a limited number of resampling methods were explored in the previous research, a small quantity of work has been done. In this study, we propose a novel illustrative study of the performance of the existing resampling techniques, viz. oversampling, undersampling, and hybrid sampling, for removing the imbalance from the minor samples of the hyperspectral dataset. The balanced dataset is classified in the next step, using the tree-based ensemble classifiers by including the spectral and spatial features. Finally, the comparative study is performed based on the statistical analysis of the outcome obtained from those classifiers that are discussed in the results section. In addition, we applied a new ensemble hybrid classifier named random rotation forest to our dataset. Three benchmark hyperspectral datasets: Indian Pines, Salinas Valley, and Pavia University, are applied for performing the experiments. We have taken precision, recall, F score, Cohen kappa, and overall accuracy as assessment metrics to evaluate our model. The obtained result shows that SMOTE, Tomek Links, and their combinations stand out to be the more optimized resampling strategies. Moreover, the ensemble classifiers such as rotation forest and random rotation ensemble provide more accuracy than others of their kind.

Deep Ensembles for Hyperspectral Image Data Classification and Unmixing

Hyperspectral images capture very detailed information about scanned objects and, hence, can be used to uncover various characteristics of the materials present in the analyzed scene. However, such image data are difficult to transfer due to their large volume, and generating new ground-truth datasets that could be utilized to train supervised learners is costly, time-consuming, very user-dependent, and often infeasible in practice. The research efforts have been focusing on developing algorithms for hyperspectral data classification and unmixing, which are two main tasks in the analysis chain of such imagery. Although in both of them, the deep learning techniques have bloomed as an extremely effective tool, designing the deep models that generalize well over the unseen data is a serious practical challenge in emerging applications. In this paper, we introduce the deep ensembles benefiting from different architectural advances of convolutional base models and suggest a new approach towards aggregating the outputs of base learners using a supervised fuser. Furthermore, we propose a model augmentation technique that allows us to synthesize new deep networks based on the original one by injecting Gaussian noise into the model’s weights. The experiments, performed for both hyperspectral data classification and unmixing, show that our deep ensembles outperform base spectral and spectral-spatial deep models and classical ensembles employing voting and averaging as a fusing scheme in both hyperspectral image analysis tasks.

원거리 초분광 데이터 분류를 위한 환경 인자의 영향성 분석

This paper presents an analysis of the environmental effects on the remote hyperspectral image data classification. Spectral reflectivity and emissivity are unique properties for classifying materials in the short-wave and mid-wave infrared bands. However, various environmental factors such as solar radiation, atmospheric transmittance, and path radiance contaminate the object signature of reflectivity and emissivity, leading to degraded classification performance. Synthetic hyperspectral data sets are generated by using the radiative transfer equation and MODTRAN. The influence of each factor is analyzed quantitatively in terms of the material classification rate. These results can be fundamental guidelines in various hyperspectral signal processing in remote outdoor environments.

Hyperspectral Image Data Classification with Refined Spectral Spatial Features Based on Stacked Autoencoder Approach

Background: Hyperspectral (HS) image data comprises of tremendous amount of spatial and spectral information which offers feature identification and classification with high accuracy. As part of the Deep Learning (DL) framework Stacked Autoencoders (SAEs) has been successfully applied for deep spectral features extraction in high dimensional data. HS deep image feature extraction becomes complex and time consuming due to the hundreds of spectral bands available in the hypercubes. Methods: The proposed method aims condense the spectral-spatial information through suitable feature extraction and feature selection methods to reduce data dimension to an appropriate scale. Further, the reduced feature set is processed by SAE for final feature representation and classification. Results: The proposed method has resulted in reduced computation time by ~ 300s and an improvement in classification accuracy by ~15% as compared to uncondensed spectral-spatial features fed directly to SAE network. Conclusion: Future research could explore the combination of most state-of-the art techniques.

Development of Fusarium head blight classification index using hyperspectral microscopy images of winter wheat spikelets

Fusarium damage in wheat reduces the quality and safety of associated food and feed products. In this study, a specific Fusarium head blight (FHB) classification index (FCI) for detection of this disease in wheat is proposed. Hyperspectral microscopy images of wheat spikelets are used as the data source. An algorithm combining the instability index (ISI) and spectral angle mapper (SAM) classifier (ISI-SAM) is used to extract four sensitive single wavelengths. Then, partial least-squares regression of the disease area ratio with simple spectral vegetation indices (SVIs) for each image is used to determine the most relevant spectral index (i.e., difference spectral index, DSI (668, 417)). On this basis, in order to develop a hyperspectral index for FHB detection, an exhaustive search for the best weighted combination of a single wavelength and DSI (668, 417) is conducted, with all possible combinations being tested. The final FCI is F C I = 0.25 × [ 2 ( R 668 − R 417 ) − R 539 ] , with an overall classification accuracy of 89.80%. The FCI is tested for its ability to detect and classify the healthy and diseased areas of wheat spikelets through comparison with six commonly used SVIs, and its disease identification accuracy is almost 30% higher than that of the best-performing SVI. The FCI is also successfully applied to classification of hyperspectral image data from wheat spikes. The FCI developed in this study constitutes a stable and feasible method of early FHB detection through spectral and low-altitude remote sensing, and will improve FHB detection, identification, and monitoring performance in precision agriculture applications.

A framework of quasiconformal mapping-based kernel machine with its application to hyperspectral remote sensing

Abstract Kernel machine is a feasible and effective nonlinear feature extraction method on data analysis, for example, hyperspectral sensing data. Kernel trick improves largely the performance of learning system including recognition, clustering, prediction through the nonlinear kernel mapping from the input data space to output data space. The performance of kernel-based system is largely influenced by the function and parameter of kernel. Optimizing only the parameters is not effective to promote the kernel-based learning system, because the data distribution is not changed only changing the kernel parameter. Moreover, no any universal single kernel is very adaptive to all applications. We present a framework of quasiconformal mapping-based kernel learning machine under single kernel and multiple kernels for hyperspectral image data classification. The performance of learning system is improved largely owing to the two facts: quasiconformal kernel structure changes the data structure in the kernel empirical space; quasiconformal multikernels characterize precisely the data for improving performance on solving complex visual learning tasks. The learning framework is applied to the hyperspectral image classification, and some experiments are implemented on two hyperspectral image databases.

Spectral–spatial hyperspectral image classification with adaptive mean filter and jump regression detection

A new remotely sensed hyperspectral image data classification algorithm which integrates the adaptive mean filter and jump regression in a variational framework is introduced. First, the adaptive mean filter is used to build the posterior probability distributions in each subpixel, and the jump detection method is then used to provide the content-aware information to adjust the smoothing extent of total variation in image discontinuous areas. Experimental results on real hyperspectral datasets show the relatively good performance of the proposed algorithm in terms of the overall accuracy, average accuracy and kappa statistic.

Classification of Hyperspectral Data Using an AdaBoostSVM Technique Applied on Band Clusters

Supervised classification of hyperspectral image data using conventional statistical classification methods is difficult because a sufficient number of training samples is often not available for the wide range of spectral bands. In addition, spectral bands are usually highly correlated and contain data redundancies because of the short spectral distance between the adjacent bands. To address these limitations, a multiple classifier system based on Adaptive Boosting (AdaBoost) is proposed and evaluated to classify hyperspectral data. In this method, the hyperspectral datasets are first split into several band clusters based on the similarities between the contiguous bands. In an AdaBoost classification system, the redundant and noninformative bands in each cluster are then removed using an optimal band selection technique. Next, a support vector machine (SVM) is applied to each refined cluster based on the classification results of previous clusters, and the results of these classifiers are fused using the weights obtained from the AdaBoost processing. Experimental results with standard hyperspectral datasets clearly demonstrate the superiority of the proposed algorithm with respect to both global and class accuracies, when compared to another ensemble classifiers such as simple majority voting and Naive Bayes to combine decisions from each cluster, a standard SVM applied on the selected bands of entire datasets and on all the spectral bands. More specifically, the proposed method performs better than other approaches, especially in datasets which contain classes with greater complexity and fewer available training samples.

Semisupervised Dual-Geometric Subspace Projection for Dimensionality Reduction of Hyperspectral Image Data

Exploring the geometric prior in the dimensionality reduction (DR) of hyperspectral image data (HID) is an important issue because it can overcome the possible overclassification of spectrally homogeneous areas in the HID classification. In this paper, the local geometric similarity of hyperspectral vectors is explored in both the manifold domain and image domain, and a semisupervised dual-geometric subspace projection (DGSP) approach is proposed for the DR of HID, by utilizing both labeled and unlabeled samples. First, the geometric information in the manifold domain is captured by a sparse coding-based geometric graph, and then, a local-consistency-constrained geometric matrix is defined to reveal the geometric structure in the image domain. Second, unlabeled samples are used to refine the geometric structure by defining a pairwise similarity matrix. Third, three scatter matrices are then derived from these similarity matrices to find the optimal subspace projection that captures the most important properties of the subspaces with respect to classification. Some experiments are taken on the airborne visible infrared imaging spectrometer (AVIRIS) HID to prove the efficiency of the proposed method.

Spectral–Spatial Classification of Hyperspectral Data Using Loopy Belief Propagation and Active Learning

In this paper, we propose a new framework for spectral-spatial classification of hyperspectral image data. The proposed approach serves as an engine in the context of which active learning algorithms can exploit both spatial and spectral information simultaneously. An important contribution of our paper is the fact that we exploit the marginal probability distribution which uses the whole information in the hyperspectral data. We learn such distributions from both the spectral and spatial information contained in the original hyperspectral data using loopy belief propagation. The adopted probabilistic model is a discriminative random field in which the association potential is a multinomial logistic regression classifier and the interaction potential is a Markov random field multilevel logistic prior. Our experimental results with hyperspectral data sets collected using the National Aeronautics and Space Administration's Airborne Visible Infrared Imaging Spectrometer and the Reflective Optics System Imaging Spectrometer system indicate that the proposed framework provides state-of-the-art performance when compared to other similar developments.

Research into a Feature Selection Method for Hyperspectral Imagery Using PSO and SVM

Classification and recognition of hyperspectral remote sensing images is not the same as that of conventional multi-spectral remote sensing images. We propose,a novel feature selection and classification method for hyperspectral images by combining the global optimization ability of particle swarm optimization (PSO) algorithm and the superior classification performance of a support vector machine (SVM). Global optimal search performance of PSO is improved by using a chaotic optimization search technique. Granularity based grid search strategy is used to optimize the SVM model parameters. Parameter optimization and classification of the SVM are addressed using the training date corre-sponding to the feature subset. A false classification rate is adopted as a fitness function. Tests of feature selection and classification are carried out on a hyperspectral data set. Classification performances are also compared among different feature extraction methods commonly used today. Results indicate that this hybrid method has a higher classification accuracy and can effectively extract optimal bands. A feasible approach is provided for feature selection and classifica-tion of hyperspectral image data.

Dimensionality reduction and classification of hyperspectral image data using sequences of extended morphological transformations

This work describes sequences of extended morphological transformations for filtering and classification of high-dimensional remotely sensed hyperspectral datasets. The proposed approaches are based on the generalization of concepts from mathematical morphology theory to multichannel imagery. A new vector organization scheme is described, and fundamental morphological vector operations are defined by extension. Extended morphological transformations, characterized by simultaneously considering the spatial and spectral information contained in hyperspectral datasets, are applied to agricultural and urban classification problems where efficacy in discriminating between subtly different ground covers is required. The methods are tested using real hyperspectral imagery collected by the National Aeronautics and Space Administration Jet Propulsion Laboratory Airborne Visible-Infrared Imaging Spectrometer and the German Aerospace Agency Digital Airborne Imaging Spectrometer (DAIS 7915). Experimental results reveal that, by designing morphological filtering methods that take into account the complementary nature of spatial and spectral information in a simultaneous manner, it is possible to alleviate the problems related to each of them when taken separately.

Efficient transmission and classification of hyperspectral image data

An extension of a newly developed cluster-space representation is applied to efficient data transmission and classification. Cluster-space classification, which is an automatic hybrid supervised and unsupervised classification procedure, can be performed in two stages. A semiproduct with low entropy is generated at the sender end. It is then transmitted to a range of users for further classification. Experiments using a HyMap dataset demonstrate the advantages in data transmission and the satisfactory classification accuracy.