Abstract:Hyperspectral imaging is one of the advanced remote sensing techniques. High dimensional nature of hyperspectral image makes its analysis complex. Various methods have been developed to reduce the dimension of hyperspectral image. Most commonly used dimension reduction technique is Principal Component Analysis (PCA), which is a feature extraction method. The main shortcoming of PCA method is that it does not consider the local structures. Folded-PCA (F-PCA) takes into account both global and local structures, … Show more
“…In [ 22 ], a folded PCA (F-PCA), in which both global and local structures were taken into account, preserved all useful properties of PCA. The work simplified the analysis of the high dimensional nature of a hyperspectral image.…”
At present, many Deep Neural Network (DNN) methods have been widely used for hyperspectral image classification. Promising classification results have been obtained by utilizing such models. However, due to the complexity and depth of the model, increasing the number of model parameters may lead to an overfitting of the model, especially when training data are insufficient. As the performance of the model mainly depends on sufficient data and a large network with reasonably optimized hyperparameters, using DNNs for classification requires better hardware conditions and sufficient training time. This paper proposes a feature fusion and multi-layered gradient boosting decision tree model (FF-DT) for hyperspectral image classification. First, we fuse extended morphology profiles (EMPs), linear multi-scale spatial characteristics, and nonlinear multi-scale spatial characteristics as final features to extract both special and spectral features. Furthermore, a multi-layered gradient boosting decision tree model is constructed for classification. We conduct experiments based on three datasets, which in this paper are referred to as the Pavia University, Indiana Pines, and Salinas datasets. It is shown that the proposed FF-DT achieves better performance in classification accuracy, training conditions, and time consumption than other current classical hyperspectral image classification methods.
“…In [ 22 ], a folded PCA (F-PCA), in which both global and local structures were taken into account, preserved all useful properties of PCA. The work simplified the analysis of the high dimensional nature of a hyperspectral image.…”
At present, many Deep Neural Network (DNN) methods have been widely used for hyperspectral image classification. Promising classification results have been obtained by utilizing such models. However, due to the complexity and depth of the model, increasing the number of model parameters may lead to an overfitting of the model, especially when training data are insufficient. As the performance of the model mainly depends on sufficient data and a large network with reasonably optimized hyperparameters, using DNNs for classification requires better hardware conditions and sufficient training time. This paper proposes a feature fusion and multi-layered gradient boosting decision tree model (FF-DT) for hyperspectral image classification. First, we fuse extended morphology profiles (EMPs), linear multi-scale spatial characteristics, and nonlinear multi-scale spatial characteristics as final features to extract both special and spectral features. Furthermore, a multi-layered gradient boosting decision tree model is constructed for classification. We conduct experiments based on three datasets, which in this paper are referred to as the Pavia University, Indiana Pines, and Salinas datasets. It is shown that the proposed FF-DT achieves better performance in classification accuracy, training conditions, and time consumption than other current classical hyperspectral image classification methods.
“…If the spatial size S is very large, the calculation of covariance matrix is difficult using PCA due to memory management issue [7]. Furthermore, PCA be unsuccessful to catch the individual contribution of each of the F bands and considers all bands of hyperspectral image (HSI) equally in covariance matrix calculation [8].…”
Section: Folded Principal Component Analysismentioning
confidence: 99%
“…In implementation of FPCA [7] each mean-adjusted spectral vector is transformed into a H×W matrix which is defined as (5) Finally, the overall covariance matrix for the whole dataset is obtained by accumulating all these partial matrices which is given by (6) The projection matrix is then computed after performing Eigen decomposition on .…”
Section: Folded Principal Component Analysismentioning
confidence: 99%
“…Different feature extraction and feature selection methos are available for feature reduction [5 ,6]. In current literature, Principal Component Analysis (PCA) is utilized for the task of feature extraction [1,7]. But PCA has some limitations as it depends on global variance overlooking low variant components which may cause loss of information [1].…”
Subspace detection of remote sensing hyperspectral image data cube has become an important area of research because of the challenges of dealing with high dimensional feature space for efficient identification of ground objects. Standard feature extraction method such as Principal Component Analysis (PCA) has several shortcomings as it depends solely on global variance of the data set generated ignoring the low variant components. In this paper these limitations are addressed and alternatively Folded-PCA (FPCA) is used for feature extraction. FPCA has some advantages over PCA as it utilizes both local and global structures of the image and requires comparatively less computational cost and memory. These properties make it suitable for feature extraction therefore our proposed method combines it with Quadratic Mutual Information (QMI) for the task of feature reduction. In this research, QMI is utilized as a means of feature selection over the new features generated from FPCA to obtain an informative subspace. The proposed method is named as (FPCA-QMI). It is tested on two hyperspectral datasets one is real mixed agricultural land and another one is an university area. Finally Kernel Support Vector Machine (KSVM) technique is applied to measure the classification accuracy of these two datasets. From the experimental analysis it is observed that the proposed method can detect effective subspace and obtains the highest accuracy of 98.0328% and 99.0431% on two real hyperspectral images which is better than the baseline approaches.
High dimensional problems are often encountered in studies related to hyperspectral data. One of the challenges that arise is how to find representations that are accurate so that important structures can be clearly easily. This study aims to process segmentation of hyperspectral image by using swarm optimization techniques. This experiments use Aviris Indian Pines hyperspectral image dataset that consist of 103 bands. The method used for segmentation image is particle swarm optimization (PSO), Darwinian particle swarm optimization (DPSO) and fractional order Darwinian particle swarm optimization (FODPSO). Before process segmentation image, the dimension of the hyperspectral image data set are first reduced by using independent component analysis (ICA) technique to get first independent component. The experimental show that FODPSO method is better than PSO and DPSO, in terms of the average CPU processing time and best fitness value. The PSNR and SSIM values when using FODPSO are better than the other two swarm optimization method. It can be concluded that FODPSO method is better in order to obtain better segmentation results compared to the previous method.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.