Temperature analysis of Coherent Anti-Stokes Raman spectra using a neural network approach

A neural network trained with clustered data has been applied to the extraction of temperature from vibrational Coherent Anti-Stokes Raman (CARS) spectra of nitrogen. CARS is a non-intrusive thermometry technique applied in practical combustors in industry. The advantages of clustering of training data over training with unprocessed calculated spectra is described. The method is applied to CARS data from an isothermal furnace and a liquid kerosene fuelled aeroengine combustor sector rig. Resulting temperatures have been compared with values extracted from the data using conventional least squares fitting and, where possible, mean temperatures measured by pyrometer and blackbody cavity probe. The main advantage of the neural network method is speed, with the potential for online temperature extraction at the spectral acquisition rate of 10 Hz using standard PC hardware.     

Canopy blockage and scattering effects on apparent soil spectral reflectance and its consequence in spectral mixture analysis of vegetated surfaces

Remote sensing technique has become the most efficient and common approach to estimate surface vegetation cover. Among various remote sensing algorithms, spectral mixture analysis (SMA) is the most common approach to obtain sub‐pixel surface coverage. In the SMA, spectral endmembers (the number of endmembers may vary), with invariant spectral reflectance across the whole image, are needed to conduct the mixture procedure. Although the nonlinear effect in quantifying vegetation spectral reflectance was noticed and sometimes addressed in the SMA analysis, the nonlinear effect in soil spectral reflectance is seldom discussed in the literature. In this paper, we investigate the effects of vegetation canopy on the inter‐canopy soil spectral reflectance via mathematical modelling and field measurements. We identify two mechanisms that lead to the difference between remotely sensed apparent soil spectral reflectance and actual soil spectral reflectance. One is a canopy blockage effect, leading to a reduced apparent soil spectral reflectance. The other is a canopy scattering effect, leading to an increased apparent soil spectral reflectance. Without correction, the first (second) mechanism causes an overestimated (underestimated) areal coverage of the low‐spectral‐reflectance endmember. The overall effect of canopy to soil, however, tends to overestimate fractional vegetation cover due to the relative significance of the canopy blockage effect, even though the two mechanisms vary with spectral wavelengths and spectral difference between different vegetation and soil. For the SMA of vegetated surface using multiple‐spectral remote sensing imagery (e.g., LandSat), it is recommended that infrared bands of low vegetation spectral reflectance (e.g. band 7) be first considered; if both visible and infrared bands are used, combination of bands 3, 4, and 5 is appropriate, while use of all six bands could overestimate fraction vegetation cover.     

输入层自构造神经网络及其在光谱分析中的应用

为了解决红外光谱定量分析中的特征提取和校正规模问题,提出了一种输入层自构造神经网络。这种网络能够利用训练数据的某些先验知识,自动选择输入层神经元的个数。在学习过程中,输入神经元个数从最小值1开始,根据网络误差的变化逐步增加,最终确定最佳神经元数量。这种网络模型将特征提取和参数学习过程融为一体,有利于提高建模效率。利用仿真红外光谱的定量分析实验表明,这种网络模型不仅能够对光谱数据实现高效率的波长选择,并具有抑制随机噪声和非线性干扰的能力。     

输入层自构造神经网络及其在光谱分析中的应用

为了解决红外光谱定量分析中的特征提取和校正规模问题，提出了一种输入层自构造神经网络。这种网络能够利用训练数据的某些先验知识，自然选择输入层神经元的个数。在学习过程中，输入神经元个数从最小值1开始，根据网络误差的变化逐步增加，最终确定最佳神经元数量。这种网络模型将特征提取和参数学习过程融一体，有利于提高建模效率。利用仿真红外光谱的定量分析实验表明，这种网络模型不仅能够对光谱数据实现高效率的波长选择，并具有抑制随机噪声和非线性干扰的能力。     

Fractional snow cover mapping through artificial neural network analysis of MODIS surface reflectance

Accurate areal measurements of snow cover extent are important for hydrological and climate modeling. The traditional method of mapping snow cover is binary where a pixel is considered either snow-covered or snow-free. Fractional snow cover (FSC) mapping can achieve a more precise estimate of areal snow cover extent by estimating the fraction of a pixel that is snow-covered. The most common snow fraction methods applied to Moderate Resolution Imaging Spectroradiometer (MODIS) images have been spectral unmixing and an empirical Normalized Difference Snow Index (NDSI). Machine learning is an alternative for estimating FSC as artificial neural networks (ANNs) have been successfully used for estimating the subpixel abundances of other surfaces. The advantages of ANNs are that they can easily incorporate auxiliary information such as land cover type and are capable of learning nonlinear relationships between surface reflectance and snow fraction. ANNs are especially applicable to mapping snow cover extent in forested areas where spatial mixing of surface components is nonlinear. This study developed a multilayer feed-forward ANN trained through backpropagation to estimate FSC using MODIS surface reflectance, NDSI, Normalized Difference Vegetation Index (NDVI) and land cover as inputs. The ANN was trained and validated with higher spatial-resolution FSC maps derived from Landsat Enhanced Thematic Mapper Plus (ETM+) binary snow cover maps. Testing of the network was accomplished over training and independent test areas. The developed network performed adequately with RMSE of 12% over training areas and slightly less accurately over the independent test scenes with RMSE of 14%. The developed ANN also compared favorably to the standard MODIS FSC product. The study also presents a comprehensive validation of the standard MODIS snow fraction product whose performance was found to be similar to that of the ANN.     

Bifurcation analysis of a cellular nonlinear network model via neural network approach

In this paper, receptor-based cellular nonlinear network model is studied. By applying neural network method, the ordinary differential equations being equivalent to the partial differential equations of the model are resulted. Also, the bifurcation analysis of the transformed system is presented. To support our theoretical results, some numerical examples are given.     

Prioritizing ocean colour channels by neural network input reflectance perturbation

The radiative transfer model Hydrolight was used to produce 18?000 artificial reflectance spectra representing case 1 and case 2 water conditions. Remote sensing reflectances were generated at the MERIS wavebands 412, 442, 490, 510, 560, 620, 665 and 682?nm from randomly generated triplet combinations of chlorophyll a, non‐chlorophyllous particles and coloured dissolved organic matter concentrations. These spectra were used to train multilayer perceptron neural network algorithms to perform the inversion from input reflectances to these three optically active substances. A method is proposed that establishes the neural network output error sensitivity towards changes in the individual input reflectance channels. From the output error produced for each reflectance change, a hypothesis about the importance of each band can be made. Results suggest a strong weight associated to the 620?nm band for the estimation of all three substances.     

Fourth derivative analysis of Red Sea coral reflectance spectra

Reflectance spectra of coral colonies and associated sand and rubble were obtained over the fringing reefs of Eilat, Israel using two GER 1500 radiometers. The overall spectral response curve of Red Sea corals displayed the same three inflection points reported for Pacific corals, with notable differences between in vitro and in situ measurements. Fourth derivative analysis of relatively pure spectra (filling the sensor's field of view) allowed for differentiation of coral and non-coral targets with 95–99% accuracy. The characteristic peaks revealed by the fourth derivative match those obtained on Hawaiian corals.     

Tool condition monitoring using reflectance of chip surface and neural network

A wide variety of tool condition monitoring techniques has been introduced in recent years. Among them, tool force monitoring, tool vibration monitoring and tool acoustics emission monitoring are the three most common indirect tool condition monitoring techniques. Using multiple intelligent sensors, these techniques are able to monitor tool condition with varying degrees of success. This paper presents a novel approach for the estimation of tool wear using the reflectance of cutting chip surface and a back propagation neural network. It postulates that the condition of a tool can be determined using the surface finish and color of a cutting chip. A series of experiments has been carried out. The experimental data obtained was used to train the back propagation neural network. Subsequently, the trained neural network was used to perform tool wear prediction. Results show that the prediction is in good agreement with the flank wear measured experimentally.     

The biochemical decomposition of slash pine needles from reflectance spectra using neural networks

An artificial neural network was used to control for leaf water absorption during the estimation of lignin-cellulose and nitrogen concentration from the reflectance spectra of fresh slash pine needles. The inputs to the neural network comprised of spectral indices based upon wavelengths at the centre or wings of known absorption features of the biochemical compounds of interest. The results indicate that the neural network provides more accurate estimates of water concentration than when using spectral indices alone. More importantly, lignincellulose concentrations were estimated with an accuracy of around 3 per cent relative to mean value. However, an accuracy of only 24 per cent relative to mean value was achieved for estimates of nitrogen concentration.     

Multiband decomposition and spectral discriminative analysis for motor imagery BCI via deep neural network

Human limb movement imagery, which can be used in limb neural disorders rehabilitation and brain-controlled external devices, has become a significant control paradigm in the domain of brain-computer interface (BCI). Although numerous pioneering studies have been devoted to motor imagery classification based on electroencephalography (EEG) signal, their performance is somewhat limited due to insufficient analysis of key effective frequency bands of EEG signals. In this paper, we propose a model of multiband decomposition and spectral discriminative analysis for motor imagery classification, which is called variational sample-long short term memory (VS-LSTM) network. Specifically, we first use a channel fusion operator to reduce the signal channels of the raw EEG signal. Then, we use the variational mode decomposition (VMD) model to decompose the EEG signal into six band-limited intrinsic mode functions (BIMFs) for further signal noise reduction. In order to select discriminative frequency bands, we calculate the sample entropy (SampEn) value of each frequency band and select the maximum value. Finally, to predict the classification of motor imagery, a LSTM model is used to predict the class of frequency band with the largest SampEn value. An open-access public data is used to evaluated the effectiveness of the proposed model. In the data, 15 subjects performed motor imagery tasks with elbow flexion / extension, forearm supination / pronation and hand open/close of right upper limb. The experiment results show that the average classification result of seven kinds of motor imagery was 76.2%, the average accuracy of motor imagery binary classification is 96.6% (imagery vs. rest), respectively, which outperforms the state-of-the-art deep learning-based models. This framework significantly improves the accuracy of motor imagery by selecting effective frequency bands. This research is very meaningful for BCIs, and it is inspiring for end-to-end learning research.     

Speaker indexing using neural network clustering of vowel spectra

Speaker indexing refers to the process of separating speakers within a recording and assigning indices to each unique speaker. This paper describes a new speaker indexing algorithm which dynamically generates and trains a neural network to model each postulated speaker found within a recording. Each neural network is trained to differentiate the vowel spectra of one specific speaker from all other speakers. A method for combining speaker indexing and other annotations of a recording in a general framework is also presented. The speaker indexing system is currently being incorporated into several application systems in the Speech Group at the MIT Media Lab.     

Estimating impervious surface distribution by spectral mixture analysis

Estimating the distribution of impervious surface, a major component of the vegetation-impervious surface-soil (V-I-S) model, is important in monitoring urban areas and understanding human activities. Besides its applications in physical geography, such as run-off models and urban change studies, maps showing impervious surface distribution are essential for estimating socio-economic factors, such as population density and social conditions. In this paper, impervious surface distribution, together with vegetation and soil cover, is estimated through a fully constrained linear spectral mixture model using Landsat Enhanced Thematic Mapper Plus (ETM+) data within the metropolitan area of Columbus, OH in the United States. Four endmembers, low albedo, high albedo, vegetation, and soil were selected to model heterogeneous urban land cover. Impervious surface fraction was estimated by analyzing low and high albedo endmembers. The estimation accuracy for impervious surface was assessed using Digital Orthophoto Quarterquadrangle (DOQQ) images. The overall root mean square (RMS) error was 10.6%, which is comparable to the digitizing errors of DOQQ images. Results indicate that impervious surface distribution can be derived from remotely sensed imagery with promising accuracy.     

A novel restoration approach for vegetation reflectance spectra at noisy bands using the principal component analysis method

ABSTRACT

Due to the signal-to-noise ratio (SNR) of sensors, as well as atmospheric absorption and illumination conditions, etc., hyperspectral data at some bands are of poor quality. Data restoration for noisy bands is important for many remote sensing applications. In this paper, we present a novel data-driven Principal Component Analysis (PCA) approach for restoring leaf reflectance spectra at noisy bands using the spectra at effective bands. The technique decomposes the leaf reflectance spectra into their principal components (PCs), selects the leading PCs that describe the most variance in the data, and restores the data from these components. First, the first 10 PCs were determined from a training dataset simulated by the leaf optical properties model (PROSPECT-5) that contained 99.998% of the total information in the 3636 training samples. Then, the performance of the PCA method for restoration of the reflectance at noisy bands was investigated using the ANGERS leaf optical properties dataset; the results showed the spectral root mean squared error (RMSE) is in the range 6.46 × 10^?4 to 6.44 × 10^?2, which is about 3 ? 34 times more accurate than the stepwise regression method and partial least squares method (PLSR) for the ANGERS dataset. The results also showed that if the noisy bands are far away from the effective bands, the accuracy of the restored leaf reflectance spectra will decrease. Thirdly, the reliability of the restored reflectance spectra for retrieving leaf biochemical contents was assessed using the ANGERS dataset and leaf optical properties dataset established by the Beijing Academy of Agriculture and Forestry Sciences (BAAFS). Three water-sensitive vegetation indices ? normalized difference water index (NDWI), normalized difference infrared index (NDII) and Datt water index (DWI), derived from the restored leaf spectra ? were employed to retrieve the equivalent water thickness (EWT). The results showed that the leaf water content can be accurately retrieved from the restored leaf reflectance spectra. In addition, the PCA method to restore vegetation spectral reflectance can be applied on canopy level as well. The results showed that the spectral root mean squared error (RMSE) is in the range 8.22 × 10^?4 to 1.87 × 10^?2. The performance of the restored canopy spectra was investigated according to the results of retrieving canopy equivalent water thickness (CEWT) using the five spectral indices NDWI, NDWI₁₃₇₀, NDWI₁₈₉₀, NDII and DWI. The results indicated that the restored canopy spectra can be used for retrieving CEWT reliably and improve the accuracy of retrieval compared to the results using original canopy reflectance spectra.     

A neural network approach to freeway network traffic control

The paper investigates the application of a feedforward neural network approach to freeway network control via variable direction recommendations at bifurcation locations. A nonlinear control problem is formulated and solved first by use of computationally expensive nonlinear optimization techniques. A feedforward neural network is then trained by optimally adjusting its weights so as to reproduce the optimal control law for a limited number of traffic scenarios. Generalisation properties of the neural network are investigated and a discussion of advantages and disadvantages compared with alternative control approaches is provided.     

Forward and inverse modelling of canopy directional reflectance using a neural network

This article explores the use of artificial neural networks for both forward and inverse canopy modelling. The forward neural modelling paradigm involved training a network for predicting the bidirectional reflectance distribution function (BRDF) of a canopy given the density of the trees, their height, crown shape, viewing, and illumination geometry. The neural network model was able to predict the BRDF of unseen canopy sites with 90% accuracy. Analysis of the signal captured by the model indicates that the canopy structural parameters, and illumination and viewing geometry, are essential for predicting the BRDF of vegetated surfaces. The inverse neural network model involved learning the underlying relationship between canopy structural parameters and their corresponding bidirectional reflectance. The inversion results show that the R2 between the network predicted canopy parameters and the actual canopy parameters was 0.85 for density and 0.75 for both the crown shape and the height parameters. The results of both forward and inverse modelling suggest that neural networks can model accurately the BRDF of vegetated canopies.     

Estimation of urban vegetation abundance by spectral mixture analysis

The spatio-temporal distribution of vegetation is a fundamental component of the urban environment that can be quantified using multispectral imagery. However, spectral heterogeneity at scales comparable to sensor resolution limits the utility of conventional hard classification methods with multispectral reflectance data in urban areas. Spectral mixture models may provide a physically based solution to the problem of spectral heterogeneity. The objective of this study is to examine the applicability of linear spectral mixture models to the estimation of urban vegetation abundance using Landsat Thematic Mapper (TM) data. The inherent dimensionality of TM imagery of the New York City area suggests that urban reflectance measurements may be described by linear mixing between high albedo, low albedo and vegetative endmembers. A three-component linear mixing model provides stable, consistent estimates of vegetation fraction for both constrained and unconstrained inversions of three different endmember ensembles. Quantitative validation using vegetation abundance measurements derived from high-resolution (2 m) aerial photography shows agreement to within fractional abundances of 0.1 for vegetation fractions greater than 0.2. In contrast to the Normalised Difference Vegetation Index (NDVI), vegetation fraction estimates provide a physically based measure of areal vegetation abundance that may be more easily translated to constraints on physical quantities such as vegetative biomass and evapotranspiration.     

Application of spectral mixture analysis for terrain evaluation studies

In this article, we describe an approach to calculate the spectral mixture within pixels and classify multispectral images. The results are compared with the classified images by traditional supervised rules such as Maximum Likelihood and appreciable results were accomplished. The method considers the number of endmembers that form the scene spectra, followed by the determination of their nature and finally the decomposition of the spectra into their endmembers. The only requirement for this method is a radiometrically corrected image because the endmembers are directly selected from the image. To make the method presented here more efficient, we propose to apply it only to the classes having low accuracy after a traditional supervised classification. Because the land cover classes in this study are related to a geomorphological terrain unit, we propose to mask the terrain units having problematic classes and decompose these units into their endmembers. A geomorphological analysis of the study area (Tonle Sap basin in Cambodia) was made to establish the relationship between land cover, landforms and soils through terrain mapping units. Then we performed a supervised classification of a Landsat Thematic Mapper (TM) image and of the same image merged with a SPOT-panchromatic (PAN) image, based on the land covers corresponding to the terrain mapping units. Then we masked a terrain unit having problematic spectral classes and applied the spectral mixture analysis which allowed an efficient separation of the land cover classes agglomerated in the preliminary classification. The result of this re-classification was re-inserted into the first classification and was compared statistically with the results obtained in the preliminary classification. We consider this procedure an efficient method to improve the results obtained from a supervised classification. The method can separate different land covers that were agglomerated in the preliminary segmentation. In our case, the classification accuracy for the terrain unit used (the fluvial terrace) increases from 62% (using only the TM bands) and 69% (using TM+ SPOT) to 83%.     

Non-linear mixture modelling without end-members using an artificial neural network

Many methods of analysing remotely sensed data assume that pixels are pure, and so a failure to accommodate mixed pixels may result in significant errors in data interpretation and analysis. The analysis of data containing a large proportion of mixed pixels may therefore benefit from the decomposition of the pixels into their component parts. Methods for unmixing the composition of pixels have been used in a range of studies and have often increased the accuracy of the analyses. However, many of the methods assume linear mixing and require end-member spectra, but mixing is often non-linear and end-member spectra are difficult to obtain. In this paper, an alternative approach to unmixing the composition of image pixels, which makes no assumptions about the nature of the mixing and does not require end-member spectra, is presented. The method is based on an artificial neural network (ANN) and shown in a case study to provide accurate estimates of sub-pixel land cover composition. The results of this case study showed that accurate estimates of the proportional cover of a class and its areal extent may be made. It was also shown that there was a tendency for the accuracy of the unmixing to increase with the complexity of the network and the intensity of training. The results indicate the potential to derive accurate information from remotely sensed data sets dominated by mixed pixels.     

A neural network approach to job-shop scheduling

A novel analog computational network is presented for solving NP-complete constraint satisfaction problems, i.e. job-shop scheduling. In contrast to most neural approaches to combinatorial optimization based on quadratic energy cost function, the authors propose to use linear cost functions. As a result, the network complexity (number of neurons and the number of resistive interconnections) grows only linearly with problem size, and large-scale implementations become possible. The proposed approach is related to the linear programming network described by D.W. Tank and J.J. Hopfield (1985), which also uses a linear cost function for a simple optimization problem. It is shown how to map a difficult constraint-satisfaction problem onto a simple neural net in which the number of neural processors equals the number of subjobs (operations) and the number of interconnections grows linearly with the total number of operations. Simulations show that the authors' approach produces better solutions than existing neural approaches to job-shop scheduling, i.e. the traveling salesman problem-type Hopfield approach and integer linear programming approach of J.P.S. Foo and Y. Takefuji (1988), in terms of the quality of the solution and the network complexity.