Spectral Descriptors Research Articles

An Application of Manifold Learning in Global Shape Descriptors

With the rapid expansion of applied 3D computational vision, shape descriptors have become increasingly important for a wide variety of applications and objects from molecules to planets. Appropriate shape descriptors are critical for accurate (and efficient) shape retrieval and 3D model classification. Several spectral-based shape descriptors have been introduced by solving various physical equations over a 3D surface model. In this paper, for the first time, we incorporate a specific manifold learning technique, introduced in statistics and machine learning, to develop a global, spectral-based shape descriptor in the computer graphics domain. The proposed descriptor utilizes the Laplacian Eigenmap technique in which the Laplacian eigenvalue problem is discretized using an exponential weighting scheme. As a result, our descriptor eliminates the limitations tied to the existing spectral descriptors, namely dependency on triangular mesh representation and high intra-class quality of 3D models. We also present a straightforward normalization method to obtain a scale-invariant and noise-resistant descriptor. The extensive experiments performed in this study using two standard 3D shape benchmarks—high-resolution TOSCA and McGill datasets—demonstrate that the present contribution provides a highly discriminative and robust shape descriptor under the presence of a high level of noise, random scale variations, and low sampling rate, in addition to the known isometric-invariance property of the Laplace–Beltrami operator. The proposed method significantly outperforms state-of-the-art spectral descriptors in shape retrieval and classification. The proposed descriptor is limited to closed manifolds due to its inherited inability to accurately handle manifolds with boundaries.

Spectral descriptors and supervised classifier for ammonium nitrate detection in landmines by nuclear quadrupole resonance.

The high specificity of Nuclear Quadrupole Resonance (NQR) makes it very suited for the detection of antipersonnel mines, where the intensity of the signal spectrum around the resonance frequency of the target substance is the standard decision parameter; however, radiofrequency interference, soil effects on the search coil, landmine size, burial depth, and target temperature affect signal intensity. To overcome this, the use of spectral descriptors and a supervised classifier are proposed in this work, where an assembly of decision trees was trained with NQR data collected on places where a target filled with ammonium nitrate was present and where it was not. A statistical test, comparing the proposed classifier and the solution based solely on the intensity of the signal spectrum, showed with significant evidence that the proposed classifier outperforms the traditional solution. A final blind experiment was conducted in a rural region of Colombia, where five landmines of different size filled with ammonium nitrate were shallowly buried in an area of 1.9 × 1.52 m, and the system with the proposed classifier detected four of them with three false alarms. This work is also novel in detecting ammonium nitrate in antipersonnel mines, which are typical in Colombia, the second most mined country in the world.

Stand-off Hyperspectral Raman Imaging and Random Decision Forest Classification: A Potent Duo for the Fast, Remote Identification of Explosives.

In this study, we present a stand-off hyperspectral Raman imager (HSRI) for the fast detection and classification of different explosives at a distance of 15 m. The hyperspectral image cube is created by using a liquid crystal tunable filter (LCTF) to select a specific Raman shift and sequentially imaging spectral images onto an intensified CCD camera. The laser beam is expanded to illuminate the field of view of the HSRI and thereby improves large area scanning of suspicious surfaces. The collected hyperspectral image cube (HSI) is evaluated and classified using a random decision forest (RDF) algorithm. The RDF is trained with a training set of mg-amounts of different explosives, i.e., TNT, RDX, PETN, NaClO3, and NH4NO3, on an artificial aluminum substrate. The resulting classification is validated, and variable importance is used to optimize the RDF using spectral descriptors, effectively reducing the dimensionality of the data set. Using the gained information, a faster acquisition and calculation mode can be designed, giving improved results in classification at a much higher repetition rate.

Towards better efficiency of interatomic linear machine learning potentials

Interatomic machine learning potentials have achieved maturity and became worthwhile alternative to conventional interatomic potentials. In this work we profile some characteristics of linear machine learning methods. Being numerically fast and easy to implement, these methods offer many advantages and appear to be very attractive for large length and time scale calculations. However, we emphasize that in order to be accurate on some target properties these methods eventually yield overfitting. This feature is rather independent of training database and descriptor accuracy. At the same time, the major weakness of these potentials, i.e., lower accuracy with respect to the kernel potentials, proves to be their strength: within the confidence limits of the potential fitting, one can rely on less accurate but faster descriptors in order to boost the numerical efficiency. Here, we propose a hybrid type of atomic descriptor that combines the original forms of radial and spectral descriptors. Flexibility in choice of mixing proportions between the two descriptors ensures a user defined control over accuracy/numerical efficiency of the resulting hybrid descriptor form. The performance and features of the above linear machine learning potentials are investigated for the interatomic interactions in metals of primary importance for fusion and fission applications, Fe and W. The suggested hybrid approach opens many avenues in the field of linear machine learning potentials that up to now are preferentially coupled with more robust and computationally expensive spectral descriptors.

REPRESENTATIONS OF DECAY IN THE WORKS OF CAT HOPE

AbstractThis article considers the ‘representation of decay’ in selected concert works by the Australian composer Cat Hope. It draws on a mixed-method research methodology, comparing the conceptual aspects of Hope's oeuvre with analyses of studio and live recordings of Hope's work and discussing how such ideas of ‘decay’ may play out in the sonic world. Two forms of spectral analysis are employed: firstly the analysis of spectral parameters roughness, noisiness, brightness, pitch, and centroid, and secondly a visualisation of the music as a spectrogram. The data for the spectral analyses are derived from Alexander Harker's spectral descriptor tools for MaxMSP which record a value for each parameter every 25 milliseconds. At times, values are normalised within a range of 0 and 1, as representative of how listeners experience parametrical changes (i.e. dynamics, in relative terms rather than absolutes in relation to other sounds in the work). Importantly, perception of noisiness is more acute at frequencies in which the auditory critical bands are wider, below 250 Hz (roughly below middle C), precisely the upper range specified by Hope to define instruments suitable for the Australian Bass Orchestra.

Adaptive Windowing Framework for Surface Electromyogram-Based Pattern Recognition System for Transradial Amputees.

Electromyogram (EMG)-based Pattern Recognition (PR) systems for upper-limb prosthesis control provide promising ways to enable an intuitive control of the prostheses with multiple degrees of freedom and fast reaction times. However, the lack of robustness of the PR systems may limit their usability. In this paper, a novel adaptive time windowing framework is proposed to enhance the performance of the PR systems by focusing on their windowing and classification steps. The proposed framework estimates the output probabilities of each class and outputs a movement only if a decision with a probability above a certain threshold is achieved. Otherwise (i.e., all probability values are below the threshold), the window size of the EMG signal increases. We demonstrate our framework utilizing EMG datasets collected from nine transradial amputees who performed nine movement classes with Time Domain Power Spectral Descriptors (TD-PSD), Wavelet and Time Domain (TD) feature extraction (FE) methods and a Linear Discriminant Analysis (LDA) classifier. Nonetheless, the concept can be applied to other types of features and classifiers. In addition, the proposed framework is validated with different movement and EMG channel combinations. The results indicate that the proposed framework works well with different FE methods and movement/channel combinations with classification error rates of approximately 13% with TD-PSD FE. Thus, we expect our proposed framework to be a straightforward, yet important, step towards the improvement of the control methods for upper-limb prostheses.

A novel hyperspectral system for high resolution imaging of ice cores: Application to light-absorbing impurities and ice structure

Paleoclimatic information recorded in polar and mountain glaciers is fundamental to study climate and environmental changes. New advanced techniques can be developed to improve the accuracy of ice core measurements, and to store the retrieved information. In this context, we developed a novel system based on a hyperspectral imaging sensor to analyse optical properties of ice cores in a non-destructive way. The system is composed of a high-precision linear stage which embeds a hyperspectral imaging sensor and a halogen light source. It was designed to collect high resolution hyperspectral images of ice cores in the 380–1000 nm spectral range with a dedicated software. The system has been tested on an ice core extracted from the Adamello glacier (Eastern Alps, Italy). Different narrow-band and broad-band spectral descriptors were computed in order to analyse the optical properties of ice and impurities enclosed in ice layers. This system can be coupled with traditional techniques applied to ice cores, while predictive algorithms can be developed to estimate key variables in ice cores at a very high spatial and spectral resolution. We believe that hyperspectral imaging offers great potential for improving the analysis of ice cores in the next years and for guaranteeing the preservation of ice core records over time.

A Novel System for Content-Based Retrieval of Single and Multi-Label High-Dimensional Remote Sensing Images

This paper presents a novel content-based remote sensing (RS) image retrieval system that consists of the following. First, an image description method that characterizes both spatial and spectral information content of RS images. Second, a supervised retrieval method that efficiently models and exploits the sparsity of RS image descriptors. The proposed image description method characterizes the spectral content by three different novel spectral descriptors that are: raw pixel values, simple bag of spectral values and the extended bag of spectral values descriptors. To model the spatial content of RS images, we consider the well-known scale invariant feature transform-based bag of visual words approach. With the conjunction of the spatial and the spectral descriptors, RS image retrieval is achieved by a novel sparse reconstruction-based RS image retrieval method. The proposed method considers a novel measure of label likelihood in the framework of sparse reconstruction-based classifiers and generalizes the original sparse classifier to the case both single-label and multi-label RS image retrieval problems. Finally, to enhance retrieval performance, we introduce a strategy to exploit the sensitivity of the sparse reconstruction-based method to different dictionary words. Experimental results obtained on two benchmark archives show the effectiveness of the proposed system.

Enhanced independent spectral histogram representations in face recognition

A spectral histogram descriptor computes a set of marginal distributions based on the filter bank’s responses, and further encodes them into the images. The encoding process for local image structure takes place during the filtering stage, whereas the encoding process of global image feature is conducted during the histogram stage. One drawback of spectral histogram descriptors is their performances will be greatly deteriorated when the filter bank’s responses are not stochastically independent. To tackle this problem, a computational technique named Enhanced Independent Spectral Histogram Feature (EISHF) is proposed. EISHF is composed of four working modules: (1) unsupervised independent filter bank responses computation, (2) binary hashing, (3) XOR bitwise operation and feature encoding, and lastly, (4) block-wise histogramming. To ensure the performance of ordinary spectral histogram descriptors, an XOR operation has been delicately adopted to increase the independency of the filter responses. Tested on three public face databases, the experimental results have substantiated the performance of EISHF in handling different kinds of facial expressions, illuminations, time spans as well as facial makeup effects.

Spectral features for audio based vehicle and engine classification

In this paper we address automatic vehicle and engine identification based on audio information. Such data depend on many factors, including vehicle type, tires, speed and its change, as well as road type. In our previous research we designed a feature set for selected vehicle classes, discriminating pairs of classes. Later, we decided to expand the feature vector and find the best feature set (mainly based on spectral descriptors), possibly representative for each investigated vehicle category, which can be applied to a bigger data set, with more classes. The experiments were performed first on on-road recordings, and then continued with test bench (dyno) recordings. The paper also shows problems related to vehicles classification, which is detailed in official documents by national authority for issues related to the national road system, but simplified for automatic identification purposes. Experiments on audio-based vehicle type and engine type identification are presented and conclusions are shown.

Microbial transformation of intracellular dissolved organic matter from Microcystis aeruginosa and its effect on the binding of pyrene under oxic and anoxic conditions.

The environmental behaviour and the bioavailability of polycyclic aromatic hydrocarbons (PAHs) are strongly affected by dissolved organic matter (DOM) in aquatic environments. In this study, we investigated the dynamics of the bioavailability and character of the intracellular DOM (IDOM) from Microcystis aeruginosa (M. aeruginosa-IDOM) during 10days oxic and anoxic incubations by spectroscopy. Subsequently, the binding affinity of pyrene with the initial/altered M. aeruginosa-IDOM was estimated by fluorescence quenching method. The incubation results indicated that changes in dissolved organic carbon (DOC) concentration and selected spectral descriptors of the M. aeruginosa-IDOM under oxic condition were different from those of anoxic condition. The microbial transformation of the M. aeruginosa-IDOM resulted in an enhancement of the organic carbon-normalized binding coefficient (K DOC) of pyrene in both oxic and anoxic treatments. Moreover, only for the oxic condition, Pearson correlation analysis demonstrated that aromaticity (specific UV absorbance at 254nm, SUVA254), humification degree (humification index, HIX) and the percent distribution of humic-like component 2 (%C2) presented significantly positive correlations with the pyrene K DOC, while the percent distribution of protein-like component 1 (%C1) exhibited a negative correlation with the K DOC. However, no significant correlation was observed between any spectral descriptor and the K DOC under anoxic condition. This result suggested that the binding affinity of pyrene may be primarily influenced by the altered M. aeruginosa-IDOM characteristics associated with the biological transformation. Hence, our results provided potential evidence for resolving the inconsistency in the relationships between DOM characteristics and the binding affinities of PAHs.

Active tuning of stroke-induced vibrations by tennis players

ABSTRACTThis paper investigates how tennis players control stroke-induced vibration. Its aim is to characterise how a tennis player deals with entering vibration waves or how he/she has the ability to finely adjust them. A specific experimental procedure was designed, based on simultaneously collecting sets of kinematic, vibration and electromyographic data during forehand strokes using various commercial rackets and stroke intensities. Using 14 expert players, a wide range of excitations at spectral and temporal levels were investigated. Energetic and spectral descriptors of stroke-induced vibration occurring at the racket handle and at the player’s wrist and elbow were computed. Results indicated that vibrational characteristics are strongly governed by grip force and to a lower extent by the racket properties. Grip force management drives the amount of energy, as well as its distribution, into the forearm. Furthermore, hand-grip can be assimilated to an adaptive filter which can significantly modify the spectral parameters propagating into the player’s upper limb. A significant outcome is that these spectral characteristics are as much dependent on the player as on the racket. This contribution opens up new perspectives in equipment manufacture by underlining the need to account for player/racket interaction in the design process.

Spectral descriptors for bulk metallic glasses based on the thermodynamics of competing crystalline phases.

Metallic glasses attract considerable interest due to their unique combination of superb properties and processability. Predicting their formation from known alloy parameters remains the major hindrance to the discovery of new systems. Here, we propose a descriptor based on the heuristics that structural and energetic ‘confusion' obstructs crystalline growth, and demonstrate its validity by experiments on two well-known glass-forming alloy systems. We then develop a robust model for predicting glass formation ability based on the geometrical and energetic features of crystalline phases calculated ab initio in the AFLOW framework. Our findings indicate that the formation of metallic glass phases could be much more common than currently thought, with more than 17% of binary alloy systems potential glass formers. Our approach pinpoints favourable compositions and demonstrates that smart descriptors, based solely on alloy properties available in online repositories, offer the sought-after key for accelerated discovery of metallic glasses.

Wavelet-based spectral descriptors for detection of damage in sunflower seeds

Analysis of seeds is essential for determining seed lot quality and hence its sowing value. Although not fully objective, assessing seed quality with radiographic images has been used as alternative to standard laboratory testing. Here, we applied 2D scale-mixing non-decimated wavelet transform for automatic processing of radiographic images of sunflower seeds. From the transformed images several spectral indices are derived. These descriptors involve spectral slopes which are directly connected with the degree of image regularity. A methodology paradigm was developed to analyze the images and classify each seed as damaged or undamaged (slight, full). By considering binary and multinomial supervised classification, the rate of correct classification was found to be 82% for damaged and full seeds, and 57% for damaged, slightly damaged, and full seeds. Although in principle many different transforms can serve as a basis in deriving spectral indices, this particular transform proved out to be sensitive to image anisotropy (by its scale-mixing nature), and stable in computation (by its non-decimated nature).

In vivo Raman Microspectroscopy: Intra- and Intersubject Variability of Stratum Corneum Spectral Markers

Background: In vivo Raman spectroscopy is a powerful tool for real-time analysis and in situ evaluation of tissues such as the skin. The efficiency of this technique has been widely demonstrated as a label-free method for in vivo evaluation of the skin. The aim of this study is to gather information about inter- and intra-individual variations in the spectral descriptors of water content and structure, organization of the lipid barrier and structure of proteins in the stratum corneum (SC). Methods: In vivo SC measurements were performed on 17 female volunteers aged 20-30 years (phototypes I and II). For intra-individual variability, spectral collection was performed on 5 successive days per volunteer. Shapiro-Wilk and Cochran tests were applied to check the normality and the homoscedasticity of variances. ANOVA was then applied to evaluate intra- and intergroup variability. Results: ANOVA was performed on the spectral descriptors of water content and structure, organization of the lipid barrier and secondary structure of proteins in the SC. No significant intra- and interday variability was observed for all volunteers. Despite the low value of the total relative standard deviation, a highly significant variation was observed between volunteers. Conclusion: Interindividual variability for Raman measurements is significant for a set of volunteers with normal nondiseased SC and close phototypes. This variability should be taken into consideration as a threshold for significant variance when working in vivo.

A spectral–structural bag-of-features scene classifier for very high spatial resolution remote sensing imagery

Land-use classification of very high spatial resolution remote sensing (VHSR) imagery is one of the most challenging tasks in the field of remote sensing image processing. However, the land-use classification is hard to be addressed by the land-cover classification techniques, due to the complexity of the land-use scenes. Scene classification is considered to be one of the expected ways to address the land-use classification issue. The commonly used scene classification methods of VHSR imagery are all derived from the computer vision community that mainly deal with terrestrial image recognition. Differing from terrestrial images, VHSR images are taken by looking down with airborne and spaceborne sensors, which leads to the distinct light conditions and spatial configuration of land cover in VHSR imagery. Considering the distinct characteristics, two questions should be answered: (1) Which type or combination of information is suitable for the VHSR imagery scene classification? (2) Which scene classification algorithm is best for VHSR imagery? In this paper, an efficient spectral–structural bag-of-features scene classifier (SSBFC) is proposed to combine the spectral and structural information of VHSR imagery. SSBFC utilizes the first- and second-order statistics (the mean and standard deviation values, MeanStd) as the statistical spectral descriptor for the spectral information of the VHSR imagery, and uses dense scale-invariant feature transform (SIFT) as the structural feature descriptor. From the experimental results, the spectral information works better than the structural information, while the combination of the spectral and structural information is better than any single type of information. Taking the characteristic of the spatial configuration into consideration, SSBFC uses the whole image scene as the scope of the pooling operator, instead of the scope generated by a spatial pyramid (SP) commonly used in terrestrial image classification. The experimental results show that the whole image as the scope of the pooling operator performs better than the scope generated by SP. In addition, SSBFC codes and pools the spectral and structural features separately to avoid mutual interruption between the spectral and structural features. The coding vectors of spectral and structural features are then concatenated into a final coding vector. Finally, SSBFC classifies the final coding vector by support vector machine (SVM) with a histogram intersection kernel (HIK). Compared with the latest scene classification methods, the experimental results with three VHSR datasets demonstrate that the proposed SSBFC performs better than the other classification methods for VHSR image scenes.

Harmonic mean normalized Laplace–Beltrami spectral descriptor

This paper proposes a framework based on harmonic mean normalized Laplace---Beltrami spectral descriptor. A series of experiments show that the harmonic mean normalization has better performance for non-rigid 3D retrieval, and it is robust to holes, local scaling, noise and sampling. To better distinguish shapes with fine or rough details, weighting method and fusion method are also employed. Weighting method reduces the negative impact of high-frequency information, and fusion method combines multi-level spectral information in both low and high frequencies. Our approach has better performance than other state-of-the-art methods on both retrieval accuracy and time consumption for stretched non-rigid 3D shapes.

A framework for live and cross platform fingerspelling recognition using modified shape matrix variants on depth silhouettes

Automatic recognition of fingerspelling postures in a live environment is a challenging task primarily due to the complex computation of popular moment-based and spectral descriptors. Shape matrix offers a time-efficient alternative that samples the shape region through the intersection points of adjacent log-polar sections. However, sparse sampling of the region by discrete log-polar intersection points cannot capture salience of the shape. This manuscript proposes modified forms of the shape matrix which can capture salience of the fingerspelling postures by the precise sampling of contours and regions. For effective segmentation and subsequent description, hand postures are acquired through the depth sensor. Proposed shape matrix variants are evaluated for fingerspelling recognition with one-handed and two-handed postures. Experiments are rigorously performed on three datasets including one-handed signs of American Sign Language (ASL), NTU hand digits, and both one-handed and two-handed signs of Indian Sign Language (ISL). Proposed shape matrix variants supersede the benchmark shape context and Gabor features by obtaining 94.15% accuracy on ISL dataset with minimum mean running time of 0.029 s. On ASL and NTU datasets, 91.86% and 95.11% accuracies are obtained with 0.0172 and 0.0483 s mean running times, respectively.

The effect of lymph drainage on the myoelectric manifestation of vastus lateralis fatigue: Preliminary results.

Variations in surface electromyograms (EMGs) collected from the vastus lateralis muscle during isometric fatiguing contractions were investigated pre-post lymphatic drainage (intervention group, N=3) and pre-post rest (control group, N=3). The slope of conduction velocity and of EMG amplitude and spectral descriptors was computed from the start to the failure time; the instant after which subjects could not endure contractions. When compared to subjects in the control group, those in the intervention group endured longer. Similarly, muscle fatigue affected to a lesser extent EMGs following lymphatic drainage than following rest. These preliminary results suggest the lymphatic drainage may potentially delay muscle fatigue.

A Local Structural Descriptor for Image Matching via Normalized Graph Laplacian Embedding.

This paper investigates graph spectral approaches to the problem of point pattern matching. Specifically, we concentrate on the issue of how to effectively use graph spectral properties to characterize point patterns in the presence of positional jitter and outliers. A novel local spectral descriptor is proposed to represent the attribute domain of feature points. For a point in a given point-set, weight graphs are constructed on its neighboring points and then their normalized Laplacian matrices are computed. According to the known spectral radius of the normalized Laplacian matrix, the distribution of the eigenvalues of these normalized Laplacian matrices is summarized as a histogram to form a descriptor. The proposed spectral descriptor is finally combined with the approximate distance order for recovering correspondences between point-sets. Extensive experiments demonstrate the effectiveness of the proposed approach and its superiority to the existing methods.