Jaime Zabalza

Novel segmented stacked autoencoder for effective dimensionality reduction and feature extraction in hyperspectral imaging

Stacked autoencoders (SAEs), as part of the deep learning (DL) framework, have been recently proposed for feature extraction in hyperspectral remote sensing. With the help of hidden nodes in deep layers, a high-level abstraction is achieved for data reduction whilst maintaining the key information of the data. As hidden nodes in SAEs have to deal simultaneously with hundreds of features from hypercubes as inputs, this increases the complexity of the process and leads to limited abstraction and performance. As such, segmented SAE (S-SAE) is proposed by confronting the original features into smaller data segments, which are separately processed by different smaller SAEs. This has resulted in reduced complexity but improved efficacy of data abstraction and accuracy of data classification.

Novel two dimensional singular spectrum analysis for effective feature extraction and data classification in hyperspectral imaging

Feature extraction is of high importance for effective data classification in hyperspectral imaging (HSI). Considering the high correlation among band images, spectral-domain feature extraction is widely employed. For effective spatial information extraction, a 2-D extension to singular spectrum analysis (2D-SSA), which is a recent technique for generic data mining and temporal signal analysis, is proposed. With 2D-SSA applied to HSI, each band image is decomposed into varying trends, oscillations, and noise. Using the trend and the selected oscillations as features, the reconstructed signal, with noise highly suppressed, becomes more robust and effective for data classification. Three publicly available data sets for HSI remote sensing data classification are used in our experiments. Comprehensive results using a support vector machine classifier have quantitatively evaluated the efficacy of the proposed approach. Benchmarked with several state-of-the-art methods including 2-D empirical mode decomposition (2D-EMD), it is found that our proposed 2D-SSA approach generates the best results in most cases. Unlike 2D-EMD that requires sequential transforms to obtain detailed decomposition, 2D-SSA extracts all components simultaneously. As a result, the execution time in feature extraction can be also dramatically reduced. The superiority in terms of enhanced discrimination ability from 2D-SSA is further validated when a relatively weak classifier, i.e., the k-nearest neighbor, is used for data classification. In addition, the combination of 2D-SSA with 1-D principal component analysis (2D-SSA-PCA) has generated the best results among several other approaches, demonstrating the great potential in combining 2D-SSA with other approaches for effective spatial-spectral feature extraction and dimension reduction in HSI.

Effective Feature Extraction and Data Reduction in Remote Sensing Using Hyperspectral Imaging [Applications Corner]

With numerous and contiguous spectral bands acquired from visible light (400- 1,000 nm) to (near) infrared (1,000-1,700 nm and over), hyperspectral imaging (HSI) can potentially identify different objects by detecting minor changes in temperature, moisture, and chemical content. As a result, HSI has been widely applied in a number of application areas, including remote sensing. HSI data contains two-dimensional (2-D) spatial and one-dimensional spectral information, and naturally forms a three-dimensional (3-D) hypercube with a high spectral resolution in nanometers that enables robust discrimination of ground features. This article discusses several variations and extensions of conventional PCA to address the aforementioned challenges. These variations and extensions include slicing the HSI data for efficient computation of the covariance matrix similarly done in 2-D-PCA analysis and grouping the spectral data to preserve the local structures and further speedup the process to determine the covariance matrix. In addition, we also discuss some non-PCA-based approaches for feature extraction and data reduction, based on techniques such as band selection, random projection, singular value decomposition, and machine-learning approaches such as the support vector machine (SVM).

Singular Spectrum Analysis for Effective Feature Extraction in Hyperspectral Imaging

As a recent approach for time series analysis, singular spectrum analysis (SSA) has been successfully applied for feature extraction in hyperspectral imaging (HSI), leading to increased accuracy in pixel-based classification tasks. However, one of the main drawbacks of conventional SSA in HSI is the extremely high computational complexity, where each pixel requires individual and complete singular value decomposition (SVD) analyses. To address this issue, a fast implementation of SSA (F-SSA) is proposed for efficient feature extraction in HSI. Rather than applying pixel-based SVD as conventional SSA does, the fast implementation only needs one SVD applied to a representative pixel, i.e., either the median or the mean spectral vector of the HSI hypercube. The result of SVD is employed as a unique transform matrix for all the pixels within the hypercube. As demonstrated in experiments using two well-known publicly available data sets, almost identical results are produced by the fast implementation in terms of accuracy of data classification, using the support vector machine (SVM) classifier. However, the overall computational complexity has been significantly reduced.

Robust PCA for micro-doppler classification using SVM on embedded systems

In this paper, a novel feature extraction technique for micro-Doppler classification and its real-time implementation using a support vector machine classifier on a low-cost, embedded digital signal processor are presented. The effectiveness of the proposed technique is improved through exploitation of the outlier rejection capabilities of robust principal component analysis (PCA) in place of classic PCA.

Structured covariance principal component analysis for real-time onsite feature extraction and dimensionality reduction in hyperspectral imaging

Presented in a three-dimensional structure called a hypercube, hyperspectral imaging suffers from a large volume of data and high computational cost for data analysis. To overcome such drawbacks, principal component analysis (PCA) has been widely applied for feature extraction and dimensionality reduction. However, a severe bottleneck is how to compute the PCA covariance matrix efficiently and avoid computational difficulties, especially when the spatial dimension of the hypercube is large. In this paper, structured covariance PCA (SC-PCA) is proposed for fast computation of the covariance matrix. In line with how spectral data is acquired in either the push-broom or tunable filter method, different implementation schemes of SC-PCA are presented. As the proposed SC-PCA can determine the covariance matrix from partial covariance matrices in parallel even without prior deduction of the mean vector, it facilitates real-time data analysis while the hypercube is acquired. This has significantly reduced the scale of required memory and also allows efficient onsite feature extraction and data reduction to benefit subsequent tasks in coding and compression, transmission, and analytics of hyperspectral data.

Effective Denoising and Classification of Hyperspectral Images Using Curvelet Transform and Singular Spectrum Analysis

Hyperspectral imaging (HSI) classification has become a popular research topic in recent years, and effective feature extraction is an important step before the classification task. Traditionally, spectral feature extraction techniques are applied to the HSI data cube directly. This paper presents a novel algorithm for HSI feature extraction by exploiting the curvelet-transformed domain via a relatively new spectral feature processing technique-singular spectrum analysis (SSA). Although the wavelet transform has been widely applied for HSI data analysis, the curvelet transform is employed in this paper since it is able to separate image geometric details and background noise effectively. Using the support vector machine classifier, experimental results have shown that features extracted by SSA on curvelet coefficients have better performance in terms of classification accuracy over features extracted on wavelet coefficients. Since the proposed approach mainly relies on SSA for feature extraction on the spectral dimension, it actually belongs to the spectral feature extraction category. Therefore, the proposed method has also been compared with some state-of-the-art spectral feature extraction techniques to show its efficacy. In addition, it has been proven that the proposed method is able to remove the undesirable artifacts introduced during the data acquisition process. By adding an extra spatial postprocessing step to the classified map achieved using the proposed approach, we have shown that the classification performance is comparable with several recent spectral-spatial classification methods.

Singular spectrum analysis for improving hyperspectral imaging based beef eating quality evaluation

Hyperspectral imaging for non-destructive beef eating quality detection is studied.Large samples were collected in abattoir production line under industry conditions.Beef tenderness and pH value were predicted using support vector machine.Singular spectrum analysis was proposed to remove instrumental noise of HSI system.Improved prediction performance was achieved by combining SSA in HSI analysis. Detecting beef eating quality in a non-destructive way has been popular in recent years. Among various non-destructive assessing methods, the feasibility of hyperspectral imaging (HSI) system was investigated in this paper. Hyperspectral images of beef samples were collected in an abattoir production line and used for predicting the beef tenderness and pH value. Support vector machine (SVM) was applied to construct the prediction equation. Before utilizing the original HSI spectral profiles directly, we propose to use singular spectrum analysis (SSA) as a pre-processing approach, where SSA has been proven to be an effective technique for time-series analysis in diverse applications. The results indicate that SSA can remove the instrumental noise of HSI system effectively and therefore improve the prediction performance.

Fast Implementation of Singular Spectrum Analysis for Effective Feature Extraction in Hyperspectral Imaging

As a recent approach for time series analysis, singular spectrum analysis (SSA) has been successfully applied for feature extraction in hyperspectral imaging (HSI), leading to increased accuracy in pixel-based classification tasks. However, one of the main drawbacks of conventional SSA in HSI is the extremely high computational complexity, where each pixel requires individual and complete singular value decomposition (SVD) analyses. To address this issue, a fast implementation of SSA (F-SSA) is proposed for efficient feature extraction in HSI. Rather than applying pixel-based SVD as conventional SSA does, the fast implementation only needs one SVD applied to a representative pixel, i.e., either the median or the mean spectral vector of the HSI hypercube. The result of SVD is employed as a unique transform matrix for all the pixels within the hypercube. As demonstrated in experiments using two well-known publicly available data sets, almost identical results are produced by the fast implementation in terms of accuracy of data classification, using the support vector machine (SVM) classifier. However, the overall computational complexity has been significantly reduced.

Embedded SVM on TMS320C6713 for signal prediction in classification and regression applications

Support Vector Machine (SVM) is a very powerful tool for signal prediction including classification and regression. With Texas Instruments TMS320C6713 DSK, an embedded SVM is implemented, where a user friendly interface is provided via peripherals like the DIPs and LEDs. The C6713 processor in combination with the SDRAM block memory can solve the complex computation that SVM requires. Also a Real-Time utilisation of the device from Matlab environment is demonstrated. An exciting application framework is finally obtained, from which some conclusions related to the implementation and final usage are derived.