Spectral Similarity Measures Research Articles

Nonlinear dimensionality reduction for efficient and effective audio similarity searching

In this paper, we address the issue of nonlinear dimensionality reduction to efficiently index spectral audio similarity measures. We propose the embedding of the spectral similarity space to a low-dimensional Euclidean space. This guarantees the triangular inequality and allows the adoption of several indexing schemes. We enlighten the advantages of the proposed indexable method against recently proposed spectral similarity measures that are also indexable. Moreover, our method compares favorably to linear dimensionality reduction methods, like multidimensional scaling (MDS). The proposed method significantly reduces the computation time during the construction process compared to any audio measure and, simultaneously, minimizes the searching cost for similar songs. To the best of our knowledge, the important issue of audio similarity measures' scalability is addressed for the first time.

THE TEXTURE CODING METHOD OF MULTISPECTRAL DATA BASED ON SPECTRAL SIMILARITY

As the traditional methods of texture extraction can not be adapted to multispectral and hyperspectral data,this paper presents a coding method of multispectral and hyperspectral data based on spectral similarity.With the concept of mapping mode of multispectral and hyperspectral texture,the new algorithm of texture extraction performs coding by the spectral similarity measure so as to reflect the overall objective texture feature.The method extends the algorithm of extracting texture to multidimensional and hyperdimensional spectral image analysis,and gives multiscale texture combination algorithm.Studies show that the coding mapping method of extracting texture is valid and effective.This paper also demonstrates that the texture feature does greatly improve the accuracy and precision of the classification.

Explicit modeling of temporal dynamics within musical signals for acoustical unit similarity

Timbre is a major cue for the human auditory system to recognize musical sounds. To describe timbre, the temporal dynamics is an important component as well as the widely used spectral envelope. In this paper, we present new temporal dynamics similarity measures, which will prove valuable for the recognition of timbral patterns. These similarity measures are evaluated, first alone, then in conjunction with spectral envelope similarity measures, for both single tones and solo recordings. Results are provided, showing that the new temporal dynamics features significantly improve timbral pattern recognition.

A REVIEW OF MINERAL SPECTRAL IDENTIFICATION METHODS AND MODELS WITH IMAGING SPECTROMETER

Mineral identification and mineral mapping constitute one of the most successful field in the application of imaging spectra. This paper has classified the spectral identification techniques and identification models of minerals into two types: one is the spectral matching of rebuilt spectral data with standard spectra or measured spectra based on spectral similarity measure, while the other is the knowledge-based or intelligent methods based on mineralogical and mineral spectral knowledge. The classification of spectral matching, spectral similarity measure, whole spectral matching, partial spectral matching, sub-pixel identification, spectral un-mixing, end-member selection and dimensional reduction are analyzed and reviewed. The existent problems and research tendency in spectral mineral mapping are also discussed. The development of the identification technology by using whole spectral region from visible to middle and thermal infrared domain seems to be one of the most important trends in spectral mineral identification and mineral mapping.

Generalization of Subpixel Analysis for Hyperspectral Data With Flexibility in Spectral Similarity Measures

Several spectral unmixing techniques have been developed for subpixel mapping using hyperspectral data in the past two decades, among which the fully constrained least squares method based on the linear spectral mixture model (LSMM) has been widely accepted. However, the shortage of this method is that the Euclidean spectral distance measure is used, and therefore, it is sensitive to the magnitude of the spectra. While other spectral matching criteria are available, such as spectral angle mapping (SAM) and spectral information divergence (SID), the current unmixing algorithm is unable to be extended to these measures. In this paper, we propose a unified subpixel mapping framework that models the unmixing process as a best match of the unknown pixel's spectrum to a weighted sum of the endmembers' spectra. We introduce sequential quadratic programming to solve the nonlinear optimization problem encountered in the implementation of this framework. The main feature of this proposed method is that it is not restricted to any particular similarity measures. Experiments were conducted with both simulated and Hyperion data. The tests demonstrated the proposed framework's advantage in accommodating various spectral similarity measures and provided performance comparisons of the Euclidean distance measure with other spectral matching criteria including SAM, spectral correlation measure, and SID.

Critical Evaluation of Product Ion Selection and Spectral Correlation Analysis for Biomarker Screening Using Targeted Peptide Multiple Reaction Monitoring

Abstract Introduction Tandem mass spectrometry (MS/MS) has emerged as a cornerstone of proteomic screens aimed at discovering putative protein biomarkers of disease with potential clinical applications. Systematic validation of lead candidates in large numbers of samples from patient cohorts remains an important challenge. One particularly promising high throughout technique is multiple reaction monitoring (MRM), a targeted form of MS/MS by which precise peptide precursor–product ion combinations, or transitions, are selectively tracked as informative probes. Despite recent progress, however, many important computational and statistical issues remain unresolved. These include the selection of an optimal set of transitions so as to achieve sufficiently high specificity and sensitivity when profiling complex biological specimens, and the corresponding generation of a suitable scoring function to reliably confirm tentative molecular identities based on noisy spectra. Methods In this study, we investigate various empirical criteria that are helpful to consider when developing and interpreting MRM-style assays based on the similarity between experimental and annotated reference spectra. We also rigorously evaluate and compare the performance of conventional spectral similarity measures, based on only a few pre-selected representative transitions, with a generic scoring metric, termed T corr, wherein a selected product ion profile is used to score spectral comparisons. Conclusions Our analyses demonstrate that T corr is potentially more suitable and effective for detecting biomarkers in complex biological mixtures than more traditional spectral library searches.

A spectral similarity measure using Bayesian statistics

A spectral similarity measure was developed that can differentiate subtle differences between two spectra. The spectra are digitalized into a vector. The difference between the two spectra is defined by a difference vector, which is one spectrum minus the other. The spectral similarity measure is transformed into a hypothesis test of the similarities and differences between the two spectra. The scalar mean of the difference vector is used as the statistical variable for the hypothesis test. A threshold for the hypothesis that the spectra are different was proposed. The Bayesian prior odds ratio was estimated from multiple spectra of the same sample. The posterior odds ratio was used to quantity the spectral similarity measure of the two spectra. Diffuse reflectance near-infrared spectra of tobacco samples of two formulations were used to demonstrate this method. The results show that this new method can detect subtle differences between the spectra.

An evaluation of wavelet-denoised hyperspectral data for remote sensing

In this paper, hyperspectral data cubes that are denoised using a wavelet transform based hybrid spatial–spectral noise-reduction (HSSNR) algorithm are evaluated based on two approaches. The first approach relies on narrow-band vegetation indices and red-edge positions, and the second approach adopts a number of spectral similarity measures. The evaluation shows that the results from the HSSNR denoising algorithm are comparable to those from existing denoising algorithms for vegetation indices and red-edge positions and superior to those from existing denoising algorithms for spectral similarity measures.

Similarity measures for spectral discrimination of salt‐affected soils

This paper illustrates a pilot study designed to examine the spectral response of soils due to salt variations. The aim of the study includes determining whether salt‐affected soils can be discriminated based on their spectral characteristics, by establishing a relationship between soil properties and soil spectra and by testing if variations in the spectra of salt‐affected soil samples are statistically significant. To answer the research questions, a laboratory experiment was designed to simulate salt transport to a column of soil in order to provide direct measurements of soil spectra and soil properties when salt concentration in a soil sample changes. The measured spectra were examined by the application of spectral matching techniques to quantify the variations and ascertain a relationship that supports the spectral identification of saline soils. The Ward's grouping method was conducted as an exploratory tool to statistically create homogeneous classes among data, which were obtained from the application of the spectral matching techniques to salt affected soil spectra. A nonparametric statistical test (Mann–Whitney U‐test) was used to determine whether the differences between the classes are statistically significant. The results of spectral matching techniques showed differences in absorption strength, absolute reflectance and spectral angle in the near and shortwave infrared regions. The results also showed significant correlations between soil electrical conductivity (EC) and spectral similarity measures, indicating that similarity between the samples' spectra decreases as the salt concentration in the soil increases. The generated clusters indicate two classes at the highest level, which were subdivided at the next level and further subdivided into multiple subclasses as the dissimilarity decreased. The spectral data were grouped into classes and were used to test the null hypothesis by applying the Mann–Whitney U‐test. The results indicate a significance level of α<0.02 between salinity classes and α<0.05 per waveband, meaning variations between the classes are higher than within each class.

The effectiveness of spectral similarity measures for the analysis of hyperspectral imagery

In geological imaging spectrometry (i.e., hyperspectral remote sensing), surface compositional information (e.g., mineralogy and subsequently chemistry) is obtained by statistical comparison (by means of spectral matching algorithms) of known field- or library spectra to unknown image spectra. Though these algorithms are readily used, little emphasis has been given to comparison of the performance of the various spectral matching algorithms. Four spectral measures are presented: three that calculate the angle (spectral angle measure, SAM), the vector distance (Euclidean distance measure, ED) or the vector cross-correlation (spectral correlation measure, SCM), between a known reference and unknown target spectrum and a fourth measure that measures the discrepancy of probability distributions between two pixel vectors (the spectral information divergence, SID). The performance of these spectral similarity measures is compared using synthetic hyperspectral and real (i.e., Airborne Visible Infrared Imaging Spectrometer, AVIRIS) hyperspectral data of a (artificial or real) hydrothermal alteration system characterised by the minerals alunite, kaolinite, montmorillonite and quartz. Two statistics are used to assess the performance of the spectral similarity measures: the probability of spectral discrimination (PSD) and the power of spectral discrimination (PWSD). The first relates to the ability of the selected set of spectral endmembers to map a target spectrum, whereas the second expresses the capability of a spectral measure to separate two classes relative to a reference class. Analysis of the synthetic data set (i.e., simulated alteration zones with crisp boundaries at 1–2nm spectral resolution) shows that (1) the SID outperforms the classical empirical spectral matching techniques (SAM, SCM and ED), (2) that SCM (SID, SAM and ED do not) exploits the overall shape of the reflectance curve and hence its outcomes are (positively and negatively) affected by the spectral range selected, (3) SAM and ED give nearly similar results and (4) for the same reason as in (2), the SCM is also more sensitive (again in positive and negative sense) to the spectral noise added. Results from the study of AVIRIS data show that SAM yields more spectral confusion (i.e., class overlap) than SID and SCM. In turn, SID is more effective in mapping the four target minerals than SCM as it clearly outperforms SCM when the target mineral coincides with the mineral phase on the ground.

Spectral similarity measure based on fuzzy feature contrast model

In the famous feature contrast model (FCM), the similarity measure is a linear combination of the common (similar) features and the distinctive (dissimilar) features. Because of the combination, FCM is better than other similarity models in explaining human perception similarity. However, the feature of FCM is binary. By defining the fuzzy feature set, FCM is extended into fuzzy feature contrast model (FFCM). In this paper, we adapt FFCM to measure spectral similarity. A spectrum is represented as a set including two subsets. The two subsets are characterized by spectral reflectance and spectral absorption, respectively. Meanwhile, the spectral reflectance and absorption are defined as the common (similar) and distinctive (dissimilar) subset in spectral set, respectively. Our spectral similarity model is expressed as a linear combination of the common subset, distinctive subset and their interaction. The difference between our model and FFCM is interaction of two subsets is defined. Moreover, kernel principal component analysis (KPCA) is used to remove the high correlation among different bands before spectral similarity measure. Experiments show that our model is effective in spectral similarity measure.

EEG coherence in post-LSD visual hallucinations

LSD use in certain individuals may result in chronic visual hallucinations, a DSM-IV syndrome known as hallucinogen persisting perception disorder (HPPD). We studied 38 HPPD subjects with a mean of 9.7 years of persistent visual hallucinations and 33 control subjects. Measures of local and medium distance EEG spectral coherence were calculated from all subjects. Coherence, a measure of spectral similarity over time, may estimate cortical coupling. In the eyes-open state in HPPD subjects, widespread reduction of coherence was noted. However, upon eye closure, the occipital region demonstrated augmented regional coherence over many frequencies but with reduced coherence of the occipital region to more distant regions. This occipital coherence increase correlated with previously reported shortened occipital visual evoked potential latency for HPPD subjects. We speculate from coherence and known clinical and psychophysical data that, in HPPD, there is widespread cortical inhibition in the eyes-opened state, but localized and isolated occipital disinhibition upon eye closure, a state known to facilitate hallucinatory experiences. An analogy is drawn to findings in the interictal and ictal epileptic focus. In HPPD, we speculate that occipital EEG hypersynchrony resulting from increased regional coherence, when coupled with relative isolation of visual cortex, especially upon eye closure, facilitates hallucinations and illusions.

Fingerprint matching of E. coli strains with matrix-assisted laser desorption/ionization time-of-flight mass spectrometry of whole cells using a modified correlation approach.

We have developed a mathematical algorithm to compare and distinguish matrix-assisted laser desorption/ionization (MALDI) mass spectra of whole bacteria cells. This fingerprint matching technique eliminates the subjectivity involved in visually comparing two spectra to determine whether they match and it provides a quantitative measure of spectral similarity. Using it, we have distinguished twenty five different strains of a single bacteria species, E. coli. Cells are grown in culture, samples are prepared, and MALDI-TOF mass spectra are recorded for each strain. Pairs of spectra are compared by a modified cross-correlation procedure. This modified approach increases the sensitivity of correlation analysis to small spectral differences. The technique can be fine-tuned by varying the number of intervals into which spectra are divided.

Direct distance measures in factor analysis spectral resolution

AbstractA factor analysis algorithm that estimates the spectra of mixture components using the set of most dissimilar rows and/or columns is described and illustrated. This algorithm uses the distance as a measure of spectral similarity and is suitable for application to a variety of the bilinear matrix‐formatted data types produced by hyphenated and multidimensional analytical techniqes. The algorithm requires that the data matrix contain at least one row or column that corresponds to the pure spectrum of each component to effect accurate spectral resolution. The performance of the method is illustrated using the resolution of excitation and emission spectra of up to four components from experimental fluorescence excitation‐emission matrices (EEMs). In the case of the EEM, characteristic bands in an emission spectrum effect resolution of the excitation spectrum of the corresponding component, while characteristic bands in an excitation spectrum lead to resolution of the corresponding emission spectrum. The use of the set of most dissimilar rows and columns to evaluate the degree of overlap in the component spectra and compare the quality of row and column solutions is also described.

Pattern recognition studies of tandem mass spectra

Principal components analysis and pattern recognition (PCAPR) techniques were applied to MS-MS spectra of fourteen organic compounds. Each spectrum was represented as a two-dimensional matrix containing information from the MS1 spectrum as well as from one, two or three MS2 spectra. The data were reduced by calculating a one-principal component model for each spectrum which explained between 86 and 99% of the variance. Each model was used to calculate each of the spectra, and residual standard deviations (R.S.D.s) were used as a measure of spectral similarity: low R.S.D.s (< 1.0) corresponding to similar spectra and higher R.S.D.s (⪢ 1.0) to dissimilar spectra. The system shows promise for use in monitoring situations in that MS-MS spectra can be efficiently reduced and stored as principal componets models and R.S.D. calculations can be used to identify a compound based on how well its spectrum is predicted by the available reference models.