Mixed Pixel Classification Research Articles

Kernel-Based Constrained Energy Minimization for Hyperspectral Mixed Pixel Classification

One fundamental task of hyperspectral imaging is spectral unmixing. In this case, the conventional pure pixel-based hyperspectral image classification (HSIC) may not work effectively for mixed pixels. This article presents a kernel-based approach to hyperspectral mixed pixel classification (HMPC) which includes two nonlinear mixed pixel classifiers, kernel constrained energy minimization (KCEM) and kernel linearly constrained minimum variance (KLCMV) to replace the widely used pure pixel-based support vector machine (SVM) classifier. Interestingly, what the binary-class and multiclass SVM classifiers are to pure pixel-based HSIC can be similarly derived for what a single-class KCEM detector and a multiclass KLCMV detector are to HMPC. In particular, the commonly used discrete classification map-based hard classification measures, average accuracy (AA) and overall accuracy (OA) for performance evaluation can be further generalized to real-valued mixed class abundance fractional map-based soft classification measures via 3-D receiver operating characteristic (3-D ROC) analysis-derived detection measures. Extensive experiments are conducted to demonstrate the utility of HMPC where KCEM/KLCMV not only significantly improve the classification performance of CEM/LCMV-based classifiers but also outperform many existing spectral-spatial classification methods.

Band Sampling of Kernel Constrained Energy Minimization Using Training Samples for Hyperspectral Mixed Pixel Classification

Constrained energy minimization (CEM) has been extended to several generalized versions, iterative CEM (ICEM), Nystrom method-based kernel CEM (NKCEM), and iterative training sampling-based NKCEM (ITS-NKCEM) for hyperspectral image classification (HSIC). Since CEM is a subpixel target detector that specifies a target signature to detect its target abundance fractions present in data samples, this article takes the advantage of CEM’s ability in subpixel detection to consider NKCEM as a nonlinear mixed pixel classifier for hyperspectral mixed pixel classification (HMPC). Recently, a new concept of band sampling (BSam) was proposed by utilizing random signal sampling derived from compressive sensing (CS) to show its performance better than band selection (BSel) for HSIC. Thus, incorporating BSam into NKCEM and ITS-NKCEM yields two new versions for HPMC, called band sampling NKCEM (BSam-NKCEM) and band sampling ITS-NKCEM (BSam-ITS-NKCEM). Interestingly, despite that BSam-ITS-NKCEM uses sampled bands as well as training samples, extensive experiments demonstrate that it can perform better than all other CEM and KCEM versions using full bands and ground truth. Specifically, it also shows that BSam-ITS-NKCEM can do better than spectral–spatial HSIC techniques using BSel.

An Iterative Mixed Pixel Classification for Brain Tissues and White Matter Hyperintensity in Magnetic Resonance Imaging

White matter hyperintensities (WMH) generally can be detected and diagnosed by magnetic resonance imaging (MRI). It has been pointed out that WMH is closely associated with stroke, cognitive impairment, dementia, and even is very relevant to the increased risk of death. This paper proposes a new iterative linearly constrained minimum variance (ILCMV) classification-based method which expands an iterative constrained energy minimization (ICEM) detection-based method developed for hyperspectral image classification. It explores the potential of ILCMV combined with different spatial filters in classification of brain normal tissues and WMH and also develops an alternative version of ILCMV, called Multi-class ICEM (MCICEM) for a comparative study. The synthetic images in BrainWeb are used for quantitative evaluation of ILCMV and the real brain MR images are used for visual assessment. The experimental results suggest that the Gaussian filter is most suitable for ILCMV and MCICEM if the computational time is factored into consideration. Otherwise, ILCMV/MCICEM combined with a Gabor filter yields the best classification. In addition, the average Dice similarity indexes (DSI) of CSF/GM/WM volume measurement produced by ILCMV method combined with Gaussian filter were 0.936/0.948/0.975 in synthetic MR images with all noise levels and were better than the results reported in the literature. ILCMV can simultaneously classifies brain normal tissues and WMH lesions in MR brain images and does better than detection of WMH alone. In addition, its computational time is also less than MCICEM. It is our belief that the proposed methodology demonstrates its promising in classification of brain tissue and WMH in MRI applications.

Iterative Target-Constrained Interference-Minimized Classifier for Hyperspectral Classification

Despite the fact that many approaches to hyperspectral image classification are reported, specifically spectral-spatial based methods, this paper presents a rather different approach from a viewpoint of mixed pixel classification, referred to iterative target-constrained interference-minimization classifier (ITCIMC), which includes an iterative Gaussian filtered feedback process to capture the spatial contextual information so as to improve hyperspectral image classification for multiple classes at one-shot operation. In order to evaluate classification performance more effectively, new performance measures other than commonly used overall accuracy (OA) are introduced, particularly, precision rate (PR), misclassification (MC) rate which have been overlooked in hyperspectral image classification. To illustrate the differences among OA, MC rate, and PR, two concepts of a priori classification and a posteriori classification are also proposed from a statistical signal processing point of view. As shown by experiments, ITCMC generally performs significantly better than the existing spectral-spatial hyperspectral image classification techniques in terms of PR and MC rate at the expense of slight loss of OA.

USE SATELLITE IMAGES AND IMPROVE THE ACCURACY OF HYPERSPECTRAL IMAGE WITH THE CLASSIFICATION

Abstract. The best technique to extract information from remotely sensed image is classification. The problem of traditional classification methods is that each pixel is assigned to a single class by presuming all pixels within the image. Mixed pixel classification or spectral unmixing, is a process that extracts the proportions of the pure components of each mixed pixel. This approach is called spectral unmixing. Hyper spectral images have higher spectral resolution than multispectral images. In this paper, pixel-based classification methods such as the spectral angle mapper, maximum likelihood classification and subpixel classification method (linear spectral unmixing) were implemented on the AVIRIS hyper spectral images. Then, pixel-based and subpixel based classification algorithms were compared. Also, the capabilities and advantages of spectral linear unmixing method were investigated. The spectral unmixing method that implemented here is an effective technique for classifying a hyperspectral image giving the classification accuracy about 89%. The results of classification when applying on the original images are not good because some of the hyperspectral image bands are subject to absorption and they contain only little signal. So it is necessary to prepare the data at the beginning of the process. The bands can be stored according to their variance. In bands with a high variance, we can distinguish the features from each other in a better mode in order to increase the accuracy of classification. Also, applying the MNF transformation on the hyperspectral images increase the individual classes accuracy of pixel based classification methods as well as unmixing method about 20 percent and 9 percent respectively.

Relationship exploration among PPI, ATGP and VCA via theoretical analysis

The principle of orthogonality using orthogonal projection OP is the key concept to develop mean squared error estimation theory in signal processing and communications. It has also found its way in a wide variety of applications in hyperspectral imaging. This paper explores close relationships among three OP-based algorithms, pixel purity index PPI, vertex component analysis VCA in endmember extraction, automatic target generation process ATGP in mixed pixel classification and target detection, etc. and further shows that they indeed can be interpreted one way or another. Specifically, VCA and ATGP are essentially the same algorithm in the sense that they both extract targets via maximal OP from a nested sequence of successive orthogonal subspace projection spaces.

Spatio-temporal pattern of urban land cover evolvement with urban renewal and expansion in Shanghai based on mixed-pixel classification for remote sensing imagery

Research into pixel unmixing in remote sensing imagery led to the development of soft classification methods. In this article, we propose a possibilistic c repulsive medoids (PCRMdd) clustering algorithm which attempts to find c repulsive medoids as a minimal solution of a particular objective function. The PCRMdd algorithm is applied to predict the proportion of each land use class within a single pixel, and generate a set of endmember fraction images. The clustering results obtained on multi-temporal Landsat Thematic Mapper (TM)/Enhanced Thematic Mapper plus (ETM+) images of Shanghai city in China reveal the spatio-temporal pattern of Shanghai land use evolvement and urban land spatial sprawl in course of urbanization from 1989 to 2002. The spatial pattern of land use transformation with urban renewal and expansion indicates the urban land use structure is gradually optimized during vigorous urban renewal and large-scale development of Pudong area, which will have an active influence on improving urban space landscape and enhancing the quality of the ecological environment. In addition, accuracy analysis demonstrates that PCRMdd represents a robust and effective tool for mixed-pixel classification on remote sensing imagery to obtain reliable soft classification results and endmember spectral information in a noisy environment.

Applications of Kalman Filtering to Single Hyperspectral Signature Analysis

Kalman filter (KF) is a widely used statistical signal processing technique for parameter estimation. Recently, a KF-based approach to linear spectral unmixing, called KF-based linear spectral unmixing (KFLU) was developed for mixed pixel classification. However, its applicability to spectral characterization for spectral estimation, identification, and quantification has not been explored. This paper presents new applications of Kalman filtering in spectral estimation, identification and abundance quantification for which three KF-based spectral characterization signal processing techniques are developed. These techniques are completely different from the KFLU in the sense that the former performs a KF across a spectral coverage wavelength by wavelength as opposed to the latter, which implements a Kalman filter pixel vector by pixel vector throughout an entire image cube. In addition, the proposed KF-based techniques do not require a linear mixture model as KFLU does. Accordingly, they are not linear spectral unmixing methods, but rather spectral signature filters operating as if they are spectral measures.

Subpixel mapping on remote sensing imagery using a prediction model combining wavelet transform and radial basis function neural network

Soft classification methods can be used for mixed-pixel classification on remote sensing imagery by estimating different land cover class fractions of every pixel. However, the spatial distribution and location of these class components within the pixel remain unknown. To map land cover at subpixel scale and increase the spatial resolution of land cover classification maps, in this paper, a prediction model combining wavelet transform and Radial Basis Functions (RBF) neural network, abbreviated as Wavelet-RBFNN, is constructed by predicting high-frequency wavelet coefficients from low-frequency coefficients at the same resolution with RBF network and taking wavelet coefficients at coarser resolution as training samples. According to different land cover class fraction images obtained from mixed-pixel classification, based on the assumption of neighborhood dependence of wavelet coefficients, subpixel mapping on remote sensing imagery can be accomplished through two steps, i.e., prediction of land cover class compositions within subpixels and hard classification. The experimental results obtained with artificial images, QuickBird image and Landsat 7 ETM+ image indicate that the subpixel mapping method proposed in this paper can successfully produce super-resolution land cover classification maps from remote sensing imagery, outperforming cubic B-spline and Kriging interpolation method in visual effect and prediction accuracy. The Wavelet-RBFNN model can also be applied to simulate higher spatial resolution image, and automatically identify and locate land cover targets at the subpixel scales, when the cost and availability of high resolution imagery prohibit its use in many areas of work.

Analysis of a shallow water environment by multispectral satellite images using a subpixel classification algorithm

Optical satellite images observed in shallow water represent a mixture of information concerning bottom type, water quality, and water depth. In this study, we extracted such information by estimating the abundance using the mixed pixel classification technique. The method is based on the Orthogonal Subspace Projection algorithm which can first eliminate unwanted information, and then match for the information in which we are interested in. Our results indicated that using the subpixel classification approach, information correlated with water depth can be extracted from optical multi-spectral images. To further test this approach, we used satellite images of Itu-Aba Island in the southern part of the South China Sea as a test image and compared with truth depth data from echo sounder sonar system for verification. The results were promising and showed that the information extracted from the satellite image corresponding to bathymetry was highly correlated to the true water depth.

Parallel processing of remotely sensed hyperspectral imagery: full‐pixel versus mixed‐pixel classification

AbstractThe rapid development of space and computer technologies allows for the possibility to store huge amounts of remotely sensed image data, collected using airborne and satellite instruments. In particular, NASA is continuously gathering high‐dimensional image data with Earth observing hyperspectral sensors such as the Jet Propulsion Laboratory's airborne visible–infrared imaging spectrometer (AVIRIS), which measures reflected radiation in hundreds of narrow spectral bands at different wavelength channels for the same area on the surface of the Earth. The development of fast techniques for transforming massive amounts of hyperspectral data into scientific understanding is critical for space‐based Earth science and planetary exploration. Despite the growing interest in hyperspectral imaging research, only a few efforts have been devoted to the design of parallel implementations in the literature, and detailed comparisons of standardized parallel hyperspectral algorithms are currently unavailable. This paper compares several existing and new parallel processing techniques for pure and mixed‐pixel classification in hyperspectral imagery. The distinction of pure versus mixed‐pixel analysis is linked to the considered application domain, and results from the very rich spectral information available from hyperspectral instruments. In some cases, such information allows image analysts to overcome the constraints imposed by limited spatial resolution. In most cases, however, the spectral bands collected by hyperspectral instruments have high statistical correlation, and efficient parallel techniques are required to reduce the dimensionality of the data while retaining the spectral information that allows for the separation of the classes. In order to address this issue, this paper also develops a new parallel feature extraction algorithm that integrates the spatial and spectral information. The proposed technique is evaluated (from the viewpoint of both classification accuracy and parallel performance) and compared with other parallel techniques for dimensionality reduction and classification in the context of three representative application case studies: urban characterization, land‐cover classification in agriculture, and mapping of geological features, using AVIRIS data sets with detailed ground‐truth. Parallel performance is assessed using Thunderhead, a massively parallel Beowulf cluster at NASA's Goddard Space Flight Center. The detailed cross‐validation of parallel algorithms conducted in this work may specifically help image analysts in selection of parallel algorithms for specific applications. Copyright © 2008 John Wiley & Sons, Ltd.

Hyperspectral Imagery Visualization Using Double Layers

Displaying the abundant information contained in a hyperspectral image is a challenging problem. Almost any visualization approach reduces the information content. However, we want to maximize the amount of object or material information presented. A visualization approach that uses classification as an intermediate step may maximize the information transfer. In our research, we are particularly interested in the display of mixed-pixel classification results, since most pixels in a remotely sensed hyperspectral image are mixed pixels. In this paper, we propose a visualization technique that employs two layers to integrate the mixture information (i.e., endmembers and their abundances) in each pixel. Images can be displayed with any desired level of details.

Mixed principal-component-analysis/independent-component-analysis transform for hyperspectral image analysis

Principal components analysis (PCA) and independent-component analysis (ICA) are widely used transforms to perform various tasks. Mixing both transforms has not been investigated. This paper develops a new transform, called the mixed PCA/ICA transform, which combines m principal components (PCs) produced by PCA and n independent components (ICs) generated by ICA to form a new set of m+n mixed components to be used for hyperspectral image analysis. Four problems need to be addressed. One is to determine the total number of components, p, needed to be generated for the mixed (m,n)-PCA/ICA transform. The second is how to combine the PCA and ICA in a single transform. Since the ICA does not prioritize its generated ICs in the same way that the PCs are ranked by the PCA using data variances, how to generate an appropriate set of n ICs becomes a third problem. Finally, the fourth problem is to decompress the compressed data based on the mixed PCA/ICA components if there is a need to reconstruct the original data. This paper solves these four problems and further conducts experiments to demonstrate the utility of the mixed PCA/ICA transform in subpixel detection and mixed pixel classification and quantification.

Fisher's linear spectral mixture analysis

Linear spectral mixture analysis (LSMA) has been widely used in subpixel analysis and mixed-pixel classification. One commonly used approach is based on either the least square error (LSE) criterion such as least squares LSMA or the signal-to-noise ratio (SNR) such as orthogonal subspace projection (OSP). Unfortunately, it is known that such criteria are not necessarily optimal for pattern classification. This paper presents a new and alternative approach to LSMA, called Fisher's LSMA (FLSMA). It extends the well-known pure-pixel-based Fisher's linear discriminant analysis to LSMA. Interestingly, what can be done for the LSMA can be also developed for the FLSMA. Of particular interest are two types of constraints imposed on the LSMA, target signature-constrained LSMA and target abundance-constrained LSMA, which can be also derived in parallel for the FLSMA, to be called feature-vector-constrained FLSMA (FVC-FLSMA) and abundance-constrained FLSMA (AC-FLSMA), respectively. Since Fisher's ratio used by the FLSMA is a more appropriate classification criterion than the LSE or SNR used for the LSMA, the FVC-FLSMA improves over the classical least squares based LSMA and SNR-based OSP in mixed-pixel classification. Similarly, the AC-FLSMA also improves abundance-constrained least squares based LSMA in quantification of abundance fractions

The contribution of the sources separation method in the decomposition of mixed pixels

In this paper, we propose to prove the importance of the application of blind sources separation methods on remote sensing data. Indeed, satellite images are represented by radiometric values where each one is considered as a mixture of different sources. The primary goal of our research is to hand back the different sources covering the scanned zone. The main constraint to restore these sources is to take our observation images as a mixture of physically independent components. In our work, the independence between the different sources is obtained through two statistical methods. The first method is based on the reduction of the spatial source correlations, and the second one is based on the joint maximization of the fourth-order cumulants. On the opposite of the original multispectral images that are represented according to correlated axes, the source images extracted from the proposed algorithms are represented according to mutually independent axes that allow each source to represent specifically a certain type of land cover. This increases the reliability of the analysis and the interpretation of the scanned zone. The source images obtained from the application of the sources separation method give a more effective representation of the information contained on the observation images. The performance of these source images is investigated through an application for the decomposition of mixed pixels. The originality of our application comes from the determination of the mixing matrix modeling the spectral endmembers based on source filters. These filters model the sensibility of each source channel according to the different spectral bands, which give an interesting information about the spectral theme represented by the corresponding source image. This application shows that the proportions of the different land cover types existing into the pixel are better estimated through the source images than through the original multispectral images. This method could offer an interesting solution to mixed-pixel classification.

A new approach to mixed pixel classification of hyperspectral imagery based on extended morphological profiles

This paper presents a new approach to the analysis of hyperspectral images, a new class of image data that is mainly used in remote sensing applications. The method is based on the generalization of concepts from mathematical morphology to multi-channel imagery. A new vector organization scheme is described, and fundamental morphological vector operations are defined by extension. Theoretical definitions of extended morphological operations are used in the formal definition of the concept of extended morphological profile, which is used for multi-scale analysis of hyperspectral data. This approach is particularly well suited for the analysis of image scenes where most of the pixels collected by the sensor are characterized by their mixed nature, i.e. they are formed by a combination of multiple underlying responses produced by spectrally distinct materials. Experimental results demonstrate the applicability of the proposed technique in mixed pixel analysis of simulated and real hyperspectral data collected by the NASA/Jet Propulsion Laboratory Airborne Visible/Infrared Imaging Spectrometer and the DLR Digital Airborne (DAIS 7915) and Reflective Optics System Imaging Spectrometers. The proposed method works effectively in the presence of noise and low spatial resolution. A quantitative and comparative performance study with regards to other standard hyperspectral analysis methodologies reveals that the combined utilization of spatial and spectral information in the proposed technique produces classification results which are superior to those found by using the spectral information alone.

A Quantitative and Comparative Analysis of Endmember Extraction Algorithms From Hyperspectral Data

Linear spectral unmixing is a commonly accepted approach to mixed-pixel classification in hyperspectral imagery. This approach involves two steps. First, to find spectrally unique signatures of pure ground components, usually known as endmembers, and, second, to express mixed pixels as linear combinations of endmember materials. Over the past years, several algorithms have been developed for autonomous and supervised endmember extraction from hyperspectral data. Due to a lack of commonly accepted data and quantitative approaches to substantiate new algorithms, available methods have not been rigorously compared by using a unified scheme. In this paper, we present a comparative study of standard endmember extraction algorithms using a custom-designed quantitative and comparative framework that involves both the spectral and spatial information. The algorithms considered in this study represent substantially different design choices. A database formed by simulated and real hyperspectral data collected by the Airborne Visible and Infrared Imaging Spectrometer (AVIRIS) is used to investigate the impact of noise, mixture complexity, and use of radiance/reflectance data on algorithm performance. The results obtained indicate that endmember selection and subsequent mixed-pixel interpretation by a linear mixture model are more successful when methods combining spatial and spectral information are applied.

Unsupervised orthogonal subspace projection approach to magnetic resonance image classification

MR images and remotely sensed images share similar image structures and characteristics because they are acquired remotely as image sequences by spectral channels at different wavelengths. As a result, techniques developed for one may be also applicable to the other. In the past, we have witnessed that some techniques that were devel- oped for magnetic resonance imaging (MRI) found great success in re- mote sensing image applications. Unfortunately, the opposite direction is yet to be investigated. In this paper, we present an application of one successful remote sensing image classification technique, called or- thogonal subspace projection (OSP), to magnetic resonance image clas- sification. Unlike classical image classification techniques, which are de- signed on a pure pixel basis, OSP is a mixed pixel classification technique that models an image pixel as a linear mixture of different material substances assumed to be present in the image data, then es- timates the abundance fraction of each individual material substance within a pixel for classification. Technically, such mixed pixel classifica- tion is performed by estimating the abundance fractions of material sub- stances resident in a pixel, rather than assigning a class label to it as usually done in pure-pixel-based classification techniques such as a minimum-distance or nearest-neighbor rule. The advantage of mixed pixel classification has been demonstrated in many applications in re- mote sensing image processing. The MRI experiments reported in this paper further show that mixed pixel classification may have advantages over the pure pixel classification. © 2002 Society of Photo-Optical Instrumentation

Target signature-constrained mixed pixel classification for hyperspectral imagery

Linear spectral mixture analysis has been widely used for subpixel detection and mixed pixel classification. When it is implemented as constrained LSMA, the constraints are generally imposed on abundance fractions in the mixture. In this paper, we consider an alternative approach, which imposes constraints on target signature vectors rather than target abundance fractions. The idea is to constrain directions of target signature vectors of interest in two different ways. One, referred to as linearly constrained minimum variance approach develops a linear filter to constrain these target signature vectors along preassigned directions using a set of specific filter gains while minimizing the filter output variance. Another, referred to as the linearly constrained discriminant analysis (LCDA), is derived from Fisher's linear discriminant analysis, but constrains the Fisher's discriminant vectors along predetermined directions to improve classification performance. Recently, Bowles et al. introduced another target signature-constrained approach, referred to as filter-vectors method, which requires a linear mixture model to implement constraints on target signature vectors. Interestingly, it turns out that the filter-vectors method can be considered as a special version of both linearly constrained minimum variance and linearly constrained discriminant analysis approaches.

Linear spectral random mixture analysis for hyperspectral imagery

Independent component analysis (ICA) has shown success in blind source separation and channel equalization. Its applications to remotely sensed images have been investigated in recent years. Linear spectral mixture analysis (LSMA) has been widely used for subpixel detection and mixed pixel classification. It models an image pixel as a linear mixture of materials present in an image where the material abundance fractions are assumed to be unknown and nonrandom parameters. This paper considers an application of ICA to the LSMA, referred to as ICA-based linear spectral random mixture analysis (LSRMA), which describes an image pixel as a random source resulting from a random composition of multiple spectral signatures of distinct materials in the image. It differs from the LSMA in that the abundance fractions of the material spectral signatures in the LSRMA are now considered to be unknown but random independent signal sources. Two major advantages result from the LSRMA. First, it does not require prior knowledge of the materials to be used in the linear mixture model, as required for the LSMA. Second, and most importantly, the LSRMA models the abundance fraction of each material spectral signature as an independent random signal source so that the spectral variability of materials can be described by their corresponding abundance fractions and captured more effectively in a stochastic manner.