Spectral Similarity Measures Research Articles

Copper Ore Identification using Spectral Similarity Measurement from Hyperion Image, Mapping of Porphyry Copper Mineralized Zone

ABSTRACT This work illustrates the effectiveness of hyperspectal image spectroscopy and lab spectroscopy in identiûcation techniques of minerals in alteration zone of ore body. The adopted procedure involves testing of Hyperion image spectra, their processing for noise, spectral matching and spectral similarity measurement with selected library spectra. Average weighted spectral similarity; visual and statistical matching techniques were used to select end-members from image spectra. Minimum noise fraction and pixel purity index technique were used to retrieve end-member spectra from hyperion image. Hyperspectral image like hyperion has the capability to deliver laboratory standard spectroscopic result. This paper illustrates the capability of hyperion image spectra in copper ore identification and mapping of chalcopyrite outcrop. A systematic approach has been made in this paper. This approach describes how image end member spectra and laboratory spectra can be co-related to fetch accurate spectral form of chalcopyrite ores. Thus, statistical and graphical comparison has been carried out between image derived end member spectra and laboratory spectra of chalcopyrite for better accuracy. The visual measurements is satisfactory, R = 0.973 for fine and 0.976 for medium grained chalcopyrite ore. Excellent statistical signiûcance levels (90–97%) are found while comparing these spectra. There are many success stories of sub-pixel and N-dimensional feature space methods to separate the hydrothermal alteration zones from iron oxide mixed ore bodies. Thus, unmixing methods are very useful for mapping of most dominating mineral parts of a pixel from hyperspectral images which have coarse spatial resolution. Finally, mapping of mineralized zone has been achieved through sub-pixel based classifiers like spectral angle mapper (SAM), constrained energy minimization (CEM) and adaptive coherence estimator (ACE) techniques.

Hyperspectral Image Segmentation via Frequency-Based Similarity for Mixed Noise Estimation

Accurate approximation of the signal-independent (SI) and signal-dependent (SD) mixed noise from hyperspectral (HS) images is a critical task for many image processing applications where the detection of homogeneous regions plays a key role. Most of the conventional methods empirically divide images into rectangular blocks and then select the homogeneous ones, but it might result in erroneous homogeneity detection, especially for highly textured HS images. To address this challenge, a superpixel segmentation algorithm is proposed in this paper, which can decompose a noisy HS image into patches that adhere to the local structures and hence persist in homogeneous characteristic. A novel spectral similarity measure is defined in the frequency domain to make the superpixel segmentation algorithm more robust to the mixed noise. Combined with an improved scatter-plot-based homogeneous superpixel selection and a multiple linear regression-based noise parameter calculation, our method can accurately estimate SD and SI noise variances from HS images with different noise conditions and various image complexities. We evaluate the proposed method with both synthetic and real Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) HS images. Experimental results demonstrate that the proposed noise estimation method outperforms the state-of-the-art methods.

Dimensionality Reduction of Hyperspectral Image with Graph-Based Discriminant Analysis Considering Spectral Similarity

Recently, graph embedding has drawn great attention for dimensionality reduction in hyperspectral imagery. For example, locality preserving projection (LPP) utilizes typical Euclidean distance in a heat kernel to create an affinity matrix and projects the high-dimensional data into a lower-dimensional space. However, the Euclidean distance is not sufficiently correlated with intrinsic spectral variation of a material, which may result in inappropriate graph representation. In this work, a graph-based discriminant analysis with spectral similarity (denoted as GDA-SS) measurement is proposed, which fully considers curves changing description among spectral bands. Experimental results based on real hyperspectral images demonstrate that the proposed method is superior to traditional methods, such as supervised LPP, and the state-of-the-art sparse graph-based discriminant analysis (SGDA).

An approach for the classification of pioneer vegetation based on species-specific phenological patterns using laboratory spectrometric measurements

This paper aims to develop a flexible decision-tree framework for the classification of nine spectrally highly similar pioneer species for monitoring heterogeneous grassland habitats, which is based on different plant phenological indicators. For typical pioneer species, laboratory spectroscopic measurements were taken. Reflectance spectra were collected to cover a complete phenological cycle. First, a combined spectral similarity measure that consists of two independent methods was applied. Second, phenological metrics were derived from time series of NDVI for each species to describe differences in annual plant development. Further, spectral vegetation indices were applied that are related to plant physiological properties. The investigated species could be grouped into three spectral separability clusters. With increasing spectral similarities, species of the particular groups can be separated into subgroups and individual species that show a similar phenological development. The remaining subgroups with the highest spectral and phenological similarities could be divided to individual species by consideration of plant physiological parameters. The results show that the combination of different spectral methods with phenological metrics enhanced the classification of pioneer vegetation and other spectrally similar species. The approach can be used as basis for a continuous monitoring of fast changing habitats.

Incorporating Spectral Similarity Into Markov Chain Geostatistical Cosimulation for Reducing Smoothing Effect in Land Cover Postclassification

Spatial statistics provides useful methods for incorporating spatial dependence into land cover classification. However, the geometric features of land cover classes are difficult to be captured by geostatistical models due to smoothing effect. The objective of this study is to incorporate spectral similarity into the Markov chain random field (MCRF) cosimulation (coMCRF) model, that is, to propose a spectral similarity-enhanced MCRF cosimulation (SS-coMCRF) model, for land cover postclassification so that postclassification will cause less geometric loss. Two mutually complementary spectral similarity measures, Jaccard index and the spectral correlation measure, were employed as a constraining factor in SS-coMCRF. One medium spatial resolution scene with a complex landscape and one very high spatial resolution scene with an urban landscape were selected for case studies. Neural network classifier and support vector machine classifier were used to conduct land cover preclassifications. Both coMCRF and SS-coMCRF were used to postprocess preclassified images based on expert-interpreted sample datasets from multiple data sources. Compared with preclassified results that depend on only spectral information of pixels, postclassifications by both models achieved similar significant improvements in overall accuracy. However, compared with coMCRF, the SS-coMCRF model apparently improved postclassified land cover patterns by effectively capturing some geometric features (e.g., boundaries and linear stripes) and more details of land cover classes. In general, incorporating spectral similarity into land cover postclassification through SS-coMCRF may contribute significantly to the “shape” or geometric accuracy of classified land cover classes.

Non-destructive Identifying Level of Browning Development in Button Mushroom (Agaricus bisporus) Using Hyperspectral Imaging Associated with Chemometrics

Analytical technology-based solutions for automation of food processing according to the quality and safety factors are of great interest to food industry. In this study, a hyperspectral imaging system with a spectral range of 380–1000 nm was used for the detection of four levels of skin browning on button mushroom. Samples were stored at standard condition (3 ± 1 °C and 92 ± 2% R.H.) to generate different levels of browning development on cap surface. After acquisition of hypercubes, the extracted spectra were pre-processed by the Savitzky-Golay and mean normalization treats. Inter- and intra-variations of different levels of browning were calculated using spectral similarity measure. The competitive adaptive reweighted sampling algorithm was applied to extract the browning-specific wavelengths, leading to the partial least square-discriminant analysis classification accuracy of 80.6 and 80.3% for calibration and testing stages, respectively. Classification maps were generated using browning-specific wavelengths and compared to the classification maps generated from the full spectral points, PCA of full spectral points, and conventional RGB imaging bands. The results illustrate that using hyperspectral imaging and chemometrics techniques for classification of different levels of browning on mushroom cap is encouraging. Within a wider context of industrial relevance, the multispectral imaging system proposed in this research has the ability to be utilized as an online and a rapid analytical tool in the mushroom processing industry.

Automatic endmember selection and nonlinear spectral unmixing of Lunar analog minerals

While the interpretation of spectral reflectance data has been widely applied to detect the presence of minerals, determining and quantifying the abundances of minerals contained by planetary surfaces is still an open problem. With this paper we address one of the two main questions arising from the spectral unmixing problem. While the mathematical mixture model has been extensively researched, considerably less work has been committed to the selection of endmembers from a possibly huge database or catalog of potential endmembers. To solve the endmember selection problem we define a new spectral similarity measure that is not purely based on the reconstruction error, i.e. the squared difference between the modeled and the measured reflectance spectrum. To select reasonable endmembers, we extend the similarity measure by adding information extracted from the spectral absorption bands. This will allow for a better separation of spectrally similar minerals. Evaluating all possible subsets of a possibly very large catalog that contain at least one endmember leads to an exponential increase in computational complexity, rendering catalogs of 20–30 endmembers impractical. To overcome this computational limitation, we propose the usage of a genetic algorithm that, while initially starting with random subsets, forms new subsets by combining the best subsets and, to some extent, does a local search around the best subsets by randomly adding a few endmembers. A Monte-Carlo simulation based on synthetic mixtures and a catalog size varying from three to eight endmembers demonstrates that the genetic algorithm is expected to require less combinations to be evaluated than an exhaustive search if the catalog comprises 10 or more endmembers. Since the genetic algorithm evaluates some combinations multiple times, we propose a simple modification and store previously evaluated endmember combinations. The resulting algorithm is shown to never require more function evaluations than a full exhaustive search and the number of required function evaluations appears to grow less than exponentially. It thus requires considerably less time than an exhaustive search because the number of function evaluations is a hardware independent measure of the computational complexity. To evaluate the spectral similarity measure, we created a spectral reflectance catalog of selected lunar analog minerals. Based on precisely prepared mixtures of two to three components, we show that the proposed spectral similarity measure selects less false endmembers from the catalog than a similarity measure that is purely based on the reconstruction error.

Pseudo-Divergence and Bidimensional Histogram of Spectral Differences for Hyperspectral Image Processing

Spectral Information Divergence (SID) was identified as the most efficient spectral similarity measure. However, we show that divergence are not adapted to direct use on spectra. Following an idea proposed by Nidamanuri, we construct a spectral pseudo-divergence based on the Kullback-Leibler divergence. This pseudo-divergence is composed of two parts: a shape and an intensity similarity measure. Consequently, bidimensional representation of spectral differences are constructed to display the histograms of similarity between a spectral reference and the spectra from a data-set or an hyperspectral image. We prove the efficiency of the spectral similarity measure and of the bidimensional histogram of spectral differences on artificial and Cultural Heritage spectral images.

Generalization of spectral fidelity with flexible measures for the sparse representation classification of hyperspectral images

Sparse representation classification (SRC) is becoming a promising tool for hyperspectral image (HSI) classification, where the Euclidean spectral distance (ESD) is widely used to reflect the fidelity between the original and reconstructed signals. In this paper, a generalized model is proposed to extend SRC by characterizing the spectral fidelity with flexible similarity measures. To validate the flexibility, several typical similarity measures—the spectral angle similarity (SAS), spectral information divergence (SID), the structural similarity index measure (SSIM), and the ESD—are included in the generalized model. Furthermore, a general solution based on a gradient descent technique is used to solve the nonlinear optimization problem formulated by the flexible similarity measures. To test the generalized model, two actual HSIs were used, and the experimental results confirm the ability of the proposed model to accommodate the various spectral similarity measures. Performance comparisons with the ESD, SAS, SID, and SSIM criteria were also conducted, and the results consistently show the advantages of the generalized model for HSI classification in terms of overall accuracy and kappa coefficient.

Application of the Frequency Spectrum to Spectral Similarity Measures

Several frequency-based spectral similarity measures, derived from commonly-used ones, are developed for hyperspectral image classification based on the frequency domain. Since the frequency spectrum (magnitude spectrum) of the original signature for each pixel from hyperspectral data can clearly reflect the spectral features of different types of land covers, we replace the original spectral signature with its frequency spectrum for calculating the existing spectral similarity measure. The frequency spectrum is symmetrical around the direct current (DC) component; thus, we take one-half of the frequency spectrum from the DC component to the highest frequency component as the input signature. Furthermore, considering the fact that the low frequencies include most of the frequency energy, we can optimize the classification result by choosing the ratio of the frequency spectrum (from the DC component to the highest frequency component) involved in the calculation. In our paper, the frequency-based measures based on the spectral gradient angle (SAM), spectral information divergence (SID), spectral correlation mapper (SCM), Euclidean distance (ED), normalized Euclidean distance (NED) and SID × sin(SAM) (SsS) measures are called the F-SAM, F-SID, F-SCM, F-ED, F-NED and F-SsS, respectively. In the experiment, three commonly-used hyperspectral remote sensing images are employed as test data. The frequency-based measures proposed here are compared to the corresponding existing ones in terms of classification accuracy. The classification results by parameter optimization are also analyzed. The results show that, although not all frequency-based spectral similarity measures are better than the original ones, some frequency-based measures, such as the F-SsS and F-SID, exhibit a relatively better performance and have more robust applications than the other spectral similarity measures.

Combine multiple mass spectral similarity measures for compound identification

Compound identification in gas chromatography-mass spectrometry GC-MS is usually achieved by comparing a query mass spectrum with reference spectral library. The rapid growing spectral library requires a more powerful spectral similarity measure to achieve the best identification performance. In this study, seven spectrum similarity measures were combined to improve the identification accuracy. To reduce the computation time, absolute value distance ABS_VD similarity measure was chosen to construct a sub-library to be searched by all similarity measures. Particle Swarm Optimisation PSO algorithm was used to first find the optimised weights for the similarity score of each similarity measure based on the training data, and then the optimised weights were applied to the test data. Simulation study using the NIST/EPA/NIH Mass Spectral Library 2005 indicates that the combination of multiple similarity measures achieves a better performance than any single similarity measure, with the identification accuracy improved by 2.2% and 1.7% for the training data and the test data, respectively.

Precision Agriculture using Advanced Remote Sensing techniques for peanut crop in Arid Land

Precision agriculture involves studying and managing crop variations within fields that can affect crop yield. In precision agriculture farmers are adapting technology and advanced remote sensing techniques with different software to relieve decision making. Hyperspectral ground measurements can be used for giving timely information about crops in specific areas and thereby providing valuable data for decision makers. In this paper field spectroscopy measurements measured by ASD field Spec4 spectroradiometer were used to monitor the spectral response and differences of peanut crop vegetation cover reflectance due to bio-physical plant variables. The results of Tukey’s HSD showed that blue, Red and NIR spectral zones are more sufficient in the monitoring differences between peanut growth stages than green, SWIR-1 and SWIR-2 spectral zones. The results of physiological spectral indices of growth stages showed significant correlations between varied classes productivity and spectral similarity measures, indicating that similarity between the samples' spectra decreases as the pigments concentration in the plant leaves increases. Furthermore, electromagnetic peanut crop mapping was successfully employed to simulate vegetation healthy effect on canopy structure and final yield.

Application of hyperspectral imaging for characterization of intramuscular fat distribution in beef

In this study, a hyperspectral imaging system in the spectral region of 400–1000nm was used for visualization and determination of intramuscular fat concentration in beef samples. Hyperspectral images were acquired for beef samples, and spectral information was then extracted from each single sample from the fat and non-fat regions. The intramuscular fat content was chemically extracted and quantified for the same samples. Chemometrics including analysis of variance (ANOVA) and spectral similarity measures involving spectral angle measure (SAM), and Euclidian distance measure (EDM) were then used to analyze the data. An ANOVA analysis indicates that the two selected spectral variables (e.g., 650.4–736.4nm) are effective to generate ratio image for visualization of the intramuscular fat distribution in beef. The spectral similarity analysis methods, which is based on the quantifying the spectral similarities by using predetermined endmember spectrum vector, provided comparable results for characterization and detection of intramuscular fat in beef. In term of overall classification accuracy, spectral similarity measure methods outperformed the ratio image of selected bands based on the result of ANOVA analysis. The results demonstrate that proposed technique has a potential for fast and nondestructive determination of intramuscular fat in beef.

Towards automated sorting of Atlantic cod (Gadus morhua) roe, milt, and liver – Spectral characterization and classification using visible and near-infrared hyperspectral imaging

Technological solutions regarding automated sorting of food according to their quality parameters are of great interest to food industry. In this regard, automated sorting of fish rest raw materials remains as one of the key challenges for the whitefish industry. Currently, the sorting of roe, milt, and liver in whitefish fisheries is done manually. Automated sorting could enable higher profitability, flexibility in production and increase the potential for high value products from roe, milt and liver that can be used for human consumption. In this study, we investigate and present a solution for classification of Atlantic cod (Gadus morhua) roe, milt and liver using visible and near-infrared hyperspectral imaging. Recognition and classification of roe, milt and liver from fractions is a prerequisite to enabling automated sorting. Hyperspectral images of cod roe, milt and liver samples were acquired in the 400–2500 nm range and specific absorption peaks were characterized. Inter- and intra-variation of the materials were calculated using spectral similarity measure. Classification models operating on one and two optimal spectral bands were developed and compared to the classification model operating on the full VIS/NIR (400–1000 nm) range. Classification sensitivity of 70% and specificity of 94% for one-band model, and 96% and 98% for two-band model (sensitivity and specificity respectively) were achieved. Generated classification maps showed that sufficient discrimination between cod liver, roe and milt can be achieved using two optimal wavelengths. Classification between roe, milt and liver is the first step towards automated sorting.

A Novel Method for Spectral Similarity Measure by Fusing Shape and Amplitude Features

Spectral similarity measure is the basis of spectral information extraction. The description of spectral features is the key to spectral similarity measure. To express the spectral shape and amplitude features reasonably, this paper presents the definition of shape and amplitude feature vector, constructs the shape feature distance vector and amplitude feature distance vector, proposes the spectral similarity measure by fusing shape and amplitude features (SAF), and discloses the relationship of fusing SAF with Euclidean distance and spectral information divergence. Different measures were tested on the basis of United States Geological Survey (USGS) mineral_beckman_430. Generally, measures by integrating SAF achieve the highest accuracy, followed by measures based on shape features and measures based on amplitude features. In measures by integrating SAF, fusing SAF shows the highest accuracy. Fusing SAF expresses the measured results with the inner product of shape and amplitude feature distance vectors, which integrate spectral shape and amplitude features well. Fusing SAF is superior to other similarity measures that integrate SAF, such as spectral similarity scale, spectral pan-similarity measure, and normalized spectral similarity score(NS 3 ).

Feature-enhanced spectral similarity measure for the analysis of hyperspectral imagery

In hyperspectral remote sensing, the surface compositional material can be identified by means of spectral matching algorithms. In many cases, the importance of each spectral band to measure spectral similarity is different, whereas the traditional spectral matching algorithms implicitly assume all wavelength-dependent absorption features are equal. This may yield an unsatisfactory performance for spectral matching. To remedy this deficiency, we propose methods called feature-enhanced spectral similarity measures. They are hybrids of the spectral matching algorithms combined with a feature-enhanced space projection, termed feature-enhanced spectral angle measure, feature-enhanced Euclidean distance measure, feature-enhanced spectral correlation measure, and feature-enhanced spectral information divergence. The proposed methods creatively project the original spectra into spectral feature-enhanced space, in which important features for measuring the spectral similarity will be increased to a high degree, whereas features of low importance will be suppressed. In order to demonstrate the effectiveness of the proposed approaches, performances are compared on real hyperspectral image data from Airborne Visible Infrared Imaging Spectrometer. The proposed methods are found to possess significant improvements over the original four spectral matching algorithms.

基于光谱相似性度量的光谱成像干扰效果定量评估

基于光谱相似性度量的光谱成像干扰效果定量评估

Coherence Function in Noisy Linear System

The coherence function provides a measure of spectral similarity of two signals, but measurement noise decreases the values of measured coherence. When the two signals are the input and output of a linear system, any system noise also decreases the measured coherence values. In digital computations, useful coherence values require some degree of averaging to increase the degrees of freedom to more than two. These fundamental issues are presented with application to system input-output coherence and two random signals with a common component. Finally, estimated coherence of the two random signals, with varying degrees of freedom, are shown with empirical adjustments that can improve the estimate of coherence. Coherence has a wide range of biomedical applications, but this article focuses on the fundamental properties of the coherence function.

Spectral Similarity Measure Using Frequency Spectrum for Hyperspectral Image Classification

A novel spectral similarity measure approach, which is named spectral frequency spectrum difference (SFSD), is proposed for hyperspectral image classification based on the frequency spectrum of spectral signature using the Fourier transform. Many important characteristics of spectral signature can be clearly reflected in the frequency spectrum. Therefore, the spectral similarity is defined as the frequency spectrum's difference between the target and reference signatures. The frequency spectrum analysis in this study suggests that the magnitude values of the first few low-frequency components for spectral signature can effectively represent the spectral similarity. To balance the difference between the low- and high-frequency components, the frequency spectrum of the target spectral signature is taken as the normalized factor in the SFSD method. Next, the U.S. Geological Survey spectral data and two hyperspectral remote sensing images were employed as test data in our validation experiments. The new SFSD proposed here was compared with the leading approaches in terms of the spectral discriminability and classification accuracy. Results show that the SFSD exhibits a relatively better performance and has more robust applications for hyperspectral image classification.

Effectiveness of Spectral Similarity Measures to Develop Precise Crop Spectra for Hyperspectral Data Analysis

Abstract. The present study was undertaken with the objective to check effectiveness of spectral similarity measures to develop precise crop spectra from the collected hyperspectral field spectra. In Multispectral and Hyperspectral remote sensing, classification of pixels is obtained by statistical comparison (by means of spectral similarity) of known field or library spectra to unknown image spectra. Though these algorithms are readily used, little emphasis has been placed on use of various spectral similarity measures to select precise crop spectra from the set of field spectra. Conventionally crop spectra are developed after rejecting outliers based only on broad-spectrum analysis. Here a successful attempt has been made to develop precise crop spectra based on spectral similarity. As unevaluated data usage leads to uncertainty in the image classification, it is very crucial to evaluate the data. Hence, notwithstanding the conventional method, the data precision has been performed effectively to serve the purpose of the present research work. The effectiveness of developed precise field spectra was evaluated by spectral discrimination measures and found higher discrimination values compared to spectra developed conventionally. Overall classification accuracy for the image classified by field spectra selected conventionally is 51.89% and 75.47% for the image classified by field spectra selected precisely based on spectral similarity. KHAT values are 0.37, 0.62 and Z values are 2.77, 9.59 for image classified using conventional and precise field spectra respectively. Reasonable higher classification accuracy, KHAT and Z values shows the possibility of a new approach for field spectra selection based on spectral similarity measure.