Component Analysis-Based Unsupervised Linear Spectral Mixture Analysis for Hyperspectral Imagery

Abstract

Two of the most challenging issues in the unsupervised linear spectral mixture analysis (ULSMA) are: 1) determining the number of signatures to form a linear mixing model; and 2) finding the signatures used to unmix data. These two issues do not occur in supervised LSMA since the target signatures are assumed to be known a priori. With recent advances in hyperspectral sensor technology, many unknown and subtle signal sources can now be uncovered and revealed and such signal sources generally cannot be identified by prior knowledge. Even when they can, the obtained knowledge may not be reliable, accurate, or complete. As a consequence, the resulting unmixed results may be misleading. This paper addresses these issues by introducing a new concept of inter-band spectral information (IBSI), which can be used to categorize signatures into background and target classes in terms of their sample spectral statistics. It then develops a component analysis (CA)-based ULSMA where two classes of signatures can be extracted directly from the data by two different CA-based transforms without requiring prior knowledge. In order to substantiate the utility of the proposed approach, synthetic images are used for experiments and real images are further used for validation

Progressive Band Processing of Linear Spectral Unmixing for Hyperspectral Imagery

“…4, the unmixed results were similar to those obtained in [31] and [32] where all abundance fractions were focused on a single mineral signature because none of the pixels in TE was pure and they all violated ASC. Since the results are very similar and many details can be found in [31] and [32], the experiments are not included.…”

“…Most recently, it has been shown in [27] and [28] that LSU could be effective even only a small number of bands are used. According to a recently developed concept, virtual dimensionality (VD) in [1], [10], and [29], if a signature used to unmix data by LSU can be effectively represented by a spectral band, VD can be further used to determine the number of signatures needed for LSU to perform effectively [30]- [32]. The issue remained is how to find appropriate signatures that correspond to these spectral bands.…”

Fully Abundance-Constrained Sequential Endmember Finding: Linear Spectral Mixture Analysis

“…As for the CE data, the number of signatures for FCLS to generate was set to 18 as suggested by Chang et al (2010Chang et al ( , 2011a, and Figs. 9.25 and 9.26 show 18 signatures found for HYDICE data by RSQ FCLS-EFA and RSC FCLS-EFA where three realizations were generated respectively.…”

Anomaly-specified virtual dimensionality

“…While the results have shown some success they also demonstrated that VD must adapt its value to various applications. For example, the number of endmember should not be the same as the number of signatures used for linear unmixing [3,[6][7]. Also, the number of dimensions required by DR is not the same as the number of bands required to be selected by BS [3,[8][9].…”

“…In order to address this issue an unsupervised algorithm must be included to be able to remove the same type of anomalies to avoid being counted multiple values of VD. This new concept has not been found in any work in [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20], specifically, [18].…”