腔体约束LIBS结合机器学习对土壤重金属元素的定量分析

土壤重金属元素含量检测及防治,对我国农业、生态环境修复具有重大意义。利用外加腔体约束结合激光诱导击穿光谱技术（LIBS）获得土壤光谱数据,采用机器学习对土壤中重金属元素Ni和Ba含量进行分析。实验设置延迟时间为0.5～5 μs,选择Ni Ⅱ 221.648 nm和Ba Ⅱ 495.709 nm作为目标研究特征谱线,计算两种LIBS条件下延迟时间对信噪比、光谱强度及增强因子的影响。结果表明,腔体约束LIBS（CC-LIBS）可以增大光谱强度及目标元素信噪比,同时随着采集延迟时间增长,等离子体数目变少,光谱强度及信噪比逐渐减小并趋于稳定;当延迟时间设置为1 μs时,CC-LIBS条件下Ni和Ba元素特征谱线信噪比达到最优,确定此时为LIBS最优实验条件。通过最优条件获取9种含Ni和Ba元素土壤样品的光谱数据,由于采集到的每组光谱信息有12 248个数据点,利用主成分分析（PCA）对CC-LIBS条件下的光谱数据降维,在保留95%以上的土壤原始信息后,选择9个主成分作为定量分析模型的输入变量,以提高模型的运算速度。采用机器学习中的Lasso,AdaBoost和Random Forest模型,对PCA降维后的光谱数据进行建模及预测,实现土壤重金属元素Ni和Ba的定量分析。结果表明,与Lasso和AdaBoost模型相比,Random Forest模型在训练集和测试集中表现出的预测性能最优。Random Forest模型下Ni元素在测试集中的R2为0.937,RMSEP为3.037;Ba元素在测试集中的相关系数R2为0.886,均方根误差RMSEP为90.515。基于腔体约束LIBS技术结合机器学习,为土壤重金属元素的高精度检测提供了技术指导。     

基于EMCCD的AOTF成像光谱仪噪声分析与信噪比研究

声光可调谐滤波器(AOTF)成像光谱仪是一种新型高光谱成像系统。针对系统存在的波段间辐射灵敏度差异较大,低照度条件下部分波段信噪较低的问题,引入了电子倍增CCD(EMCCD)作为感光器件。对该原理仪器在普通和电子倍增两种工作模式下的噪声进行了分析,推导了信噪比模型,并利用研制的原理样机进行了实验验证。在此基础上,提出了电子倍增模式下系统动态范围的评价方法,给出了以入瞳光谱辐亮度为判据的工作模式选择方法。结果表明,信噪比模型与实测结果吻合,电子倍增模式的合理选择有效提高了低照度条件下的信噪比,改善波段间辐射灵敏度的非一致性。     

白矮主序双星光谱的卷积特征提取

通过卷积运算提取白矮主序双星的光谱特征是提高识别精度的有效手段。通过设计一维卷积神经网络,以判别学习的方式从大量混合光谱中拟合出具有稳定分布的12个卷积核,有效提取白矮主序双星的卷积特征。通过引入相对松弛的光谱类别先验分布,提出反贝叶斯学习策略以解决由于光谱抽样有偏带来的问题,显著提高识别精度。通过比较光谱在不同信噪比下的交叉熵测试误差,分析卷积特征的提取过程对光谱信噪比的鲁棒性。实验发现,基于反贝叶斯学习策略的一维卷积神经网络对白矮主序双星的识别准确率达到99.0(±0.3),超过了经典的PCA+SVM模型。卷积特征谱的池化过程以降低光谱分辨率的形式缓解了光谱噪声对识别精度的影响。当信噪比小于3时,必须通过增加模型在光谱上的迭代次数以形成稳定的卷积核;当信噪比介于3与6之间时,光谱卷积特征较为稳定;当信噪比大于6时,光谱卷积特征的稳定性显著上升,信噪比对于模型识别精度带来的影响可以忽略。     

Accuracy of spectrum estimate in fluorescence spectral microscopy with spectral filters

We establish the accuracy of the spectrum that is estimated with an inexpensive fluorescence spectral microscope utilizing a small set of spectral filters [Soriano et al, Opt. Exp. 10, 1458–1464 (2002)]. The spectrum at an arbitrary image location of the fluorescent sample is estimated as a linear superposition of basis spectra that are derived by singular value decomposition (SVD) or principal component analysis (PCA) from a spectral library of fluorescence spectra. Estimation performance is analyzed as a function of library statistics, filter selection and sequencing, minimum negativity constraint and signal to noise ratio (SNR) of fluorescence image. We consider image SNR degradations that arise from weakening of image intensity, additive Gaussian noise, intensity-dependent Poisson noise and quantization errors. The recovery of specific spectral features like spectral resolution, general similarity and peak alignments, is assessed using Linfoot’s criteria of fidelity, structural content, and correlation. We found that estimation with SVD basis spectra is more robust against image noise than that with PCA basis spectra. However for high SNR images, accurate estimation is achieved more quickly with PCA basis spectra and with better response to the application of minimum negativity constraint.     

Effects of apodization smoothing and denoising on spectral fitting

PurposeVisual review of individual spectra in magnetic resonance spectroscopic imaging (MRSI) data benefits from the application of spectral smoothing; however, if this processing step is applied prior to spectral analysis this can impact the accuracy of the quantitation. This study aims to analyze the effect of spectral denoising and apodization smoothing on the quantitation of whole-brain MRSI data obtained at short TE.MethodsShort-TE MRSI data obtained at 3 T were analyzed with no spectral smoothing, following (i) Gaussian apodization with values of 1, 2, 4, 6, and 8 Hz, and (ii) denoising using principal component analysis (dnPCA) with 3 different values for the number of retained principal components. The mean lobar white matter estimates for four metabolites, signal-to-noise ratio (SNR), spectral linewidth, and confidence intervals were compared to data reconstructed using no smoothing. Additionally, a voxel-wise comparison for N-acetylaspartate quantitation with different smoothing schemes was performed.ResultsSignificant pairwise differences were seen for all Gaussian smoothing methods as compared to no smoothing (p<0.001) in linewidth and metabolite estimates, whereas dnPCA methods showing no statistically significant differences in these measures. Confidence intervals decreased, and SNR increased with increasing levels of apodization smoothing or dnPCA denoising.ConclusionMild Gaussian apodization (≤2 Hz at 3 T) can be applied with minimal (1%) errors in quantitation; however, smoothing values greater than that can significantly affect metabolite quantification. In contrast, mild to moderate dnPCA based denoising provides quantitative results that are consistent with the analysis of unsmoothed data and this method is recommended for spectral denoising.     

噪声对单耳语言可懂度的影响

单耳通信时,周边噪声对语言可懂度产生影响。针对信号侧语音信号强度70dB时,研究3种不同类型噪声下,干扰侧不同强度噪声和信号侧不同信噪比情况的语言可懂度。实验结果表明:当信号侧信噪比大于某一阈值时,干扰侧噪声对可懂度不产生显著影响,该阈值同噪声类型有关;而在信号侧低信噪比的情形下,干扰侧适当强度噪声可提高信号侧语言可懂度,最佳干扰噪声级为78—82dB,过大的干扰侧噪声级导致可懂度下降。基于心理声学和生理学的初步机理发现:噪声环境下的语音识别中,对侧耳中耳肌肉伸缩对噪声感知的抑制提高了信号侧语言可懂度。      

基于核岭回归方法的恒星大气物理参数的自动测量

我国大科学工程项目LAMOST巡天计划每观测夜能获取多达数万条天体光谱数据,天文学家通过对天体光谱的分析观察可以获取有效的天文信息用于天文学或天体物理学的研究。而针对海量数据,寻找自动方法分析天体光谱并进行天体各种物理参数的测量就具有重要研究意义和价值。这一课题也吸引了许多学者进行研究,但目前所尝试的算法和相应结果仍然需要进一步改进,针对这一需求深入研究了核岭回归(KRR)方法在恒星大气物理参数(包括有效温度、表面重力和金属丰度)自动测量方面的应用,特别是在我国大科学工程项目LAMOST所释放光谱数据上的应用。核岭回归是岭回归算法的进一步发展,而岭回归是最小二乘方法的一种变形,其具有解决高维多重共线性问题的能力。所以KRR方法适合于处理高维的天体光谱信息,从LAMOST的第五期释放数据中随机选择了2万条被识别为恒星的光谱数据用于实验测试,该数据既包含低信噪比数据,也包含高信噪比数据(g,r,i波段平均信噪比最低至6.7,最高到793)。首先,本文对光谱进行预处理,包括三个步骤:(1)利用小波变换对光谱数据进行去噪处理;(2)因为LAMOST采用的是后期修正的流量定标设计,所以还通过流量归一化来避免部分光谱流量值不准确的问题;(3)由于每条光谱维数高达数千维,利用主成分分析方法(PCA)对光谱进行了降维。然后,利用KRR方法建立了光谱数据和标准化后的三大参数值之间的回归模型。最后,通过设计进行不同的组合实验对KRR算法模型进行了测试分析,并与经典算法支持向量回归(SVR)进行了对比。综合所有实验结果显示KRR方法对应的有效温度、表面重力和金属丰度的测试平均绝对误差分别为82.9897 K,0.1858 dex和0.1211 dex,优于SVR的144.2308 K,0.1886 dex和0.1246 dex。特别是KRR在温度测试结果上有较大优势,由此表明KRR方法能够有效地应用于天体光谱特别是恒星光谱参数的自动测量处理中。     

反射光谱特征的土壤分类模型

土壤反射光谱综合反映了土壤的理化性质和内部结构,高光谱遥感已被用于基于土壤反射光谱特性的土壤分类。已有研究一般利用土壤反射光谱一阶微分主成分作为输入量进行光谱分类模型构建,但主成分数据缺乏物理意义,且缺乏对比性、适用范围也有限。与反射率一阶微分数据相比,基于去包络线提取具有明确物理意义的特征参数,能够提高土壤分类的精度,并寻找到一种高精度土壤分类模型。选取吉林省农安县的四种典型土壤（风砂土、草甸土、黑土、黑钙土）,将采集后的土壤样本进行风干、研磨、过2 mm筛处理,采用ASD FiledSpec®3便携式光谱仪对处理后的土壤样本的可见光近红外光谱区进行测试,从而获得土壤样本的光谱数据。对光谱数据进行九点平滑、10 nm重采样处理进行降噪,将处理后的数据分别进行一阶微分主成分以及去包络线处理。利用土壤样本的去包络线提取光谱特征参数。以一阶微分主成分数据和光谱特征参数为输入量分别代入Logistic聚类模型（LR）、人工神经网络聚类模型（ANN）、K-均值聚类模型（K-means）。首先明确了不同土类之间的反射光谱曲线、去包络线的差异大小,以及相同土壤的反射率曲线、去包络线进行土壤分类的优劣,并且在去包络线的基础上提取能够区分不同土类的光谱特征参数;其次,比较一阶微分主成分与光谱特征参数作为输入量时,三种光谱分类模型精度差异并分析不同模型精度差异的原因。结果表明：（1）四种土壤的反射光谱曲线差异较小,去包络线可以极大的增强四种土壤在430～1 210 nm之间的光谱差异,并在去包络线的基础上构建具有明确物理意义的光谱特征参数。（2）将一阶微分主成分和光谱特征参数分别代入三种聚类模型可知,以光谱特征参数为输入量的土壤光谱分类模型均超过了以一阶微分主成分为输入量的模型精度,由于光谱特征参数保留了原数据的物理意义、更准确的体现了不同土壤类型之间的差异性,而一阶微分主成分数据带有一定的模糊性不同范围之间缺乏对比性,在土壤分类中以光谱特征参数作为输入量更具有优势。（3）在三类土壤分类模型中,LR的分类精度最高为76.67%,Kappa系数为0.56;ANN的分类精度中等为72.50%,Kappa系数为0.48;K-means的分类精度最低,只有65.00%,Kappa系数为0.33。研究成果可为土壤精细制图、以及土壤分类仪器的研制提供技术支持。     

Fourier-transform infrared derivative spectroscopy with an improved signal-to-noise ratio

Infrared derivative spectroscopy is a useful technique for finding peaks hidden in broad spectral features. A data acquisition technique is shown that will improve the signal-to-noise ratio (SNR) of Fourier-transform infrared (FTIR) derivative spectroscopy. Typically, in a FTIR measurement one samples each point for the same time interval. The effect of using a graded time interval is studied. The simulations presented show that the SNR of first-derivative FTIR spectroscopy will improve by 15% and that the SNR of second-derivative FTIR will improve by 34%.     

Improvement of transmission performance of adaptively modulated optical OFDM signals

An adaptively modulated optical orthogonal frequency division multiplexing (OOFDM) based system has been simulated in present research. Performance analysis in terms of bit error rate (BER) and signal to noise ratio (SNR) has been carried over the system using fixed modulation for different M-array quadrature amplitude modulation (M-QAM). An adaptive modulation over OOFDM has been analyzed over three different algorithmic schemes. Impact of adaptive modulation on spectral efficiency has also been investigated for OOFDM. Simulation results show the enhancement in terms of spectral efficiency and capacity for adaptively modulated OOFDM system.     

基于PCA-MLR和PCA-BPN的莱州湾南岸滨海平原土壤有机质高光谱预测研究

土壤有机质(SOM)含量是衡量土壤质量高低的重要指标,可以用高光谱快速测定。在以往研究中,估算模型多以特征波段与线性经验模型为基础进行构建,较少考虑波段间信息冗余和共线性,预测效果不很理想并难以进行推广。为最大化消除波段信息噪声,提高模型预测精度,选取莱州湾南岸滨海平原为研究区,系统采集了111个土壤样本和实测高光谱数据(325～1 075 nm),并测试了土壤样本的有机质含量作为因变量;通过主成分分析(PCA)将实测光谱信息降维为6个主成分,并提取水分、植被光谱特征指数(DI),以此作为自变量;最后建立多元逐步线性回归(MLR)和BP神经网络(BPN)预测模型,分析不同模型对土壤有机质预测的效果。结果表明：①经过主成分的波段信息分析判别提取出6个主成分,可以表征叶绿素残留物、盐分、腐殖酸、物化矿渣和微地貌的光谱特征。②基于6个主成分作为自变量所建立的BPN模型预测精度优于MLR模型,他们的R2分别为0.704和0.643。将水分和植被光谱特征指数作为自变量增加到预测模型后,MLR和BPN的预测精度分别提高了6.1%和5.2%,R2达到0.712和0.764;③将光谱主成分和光谱特征指数作为自变量的BPN模型进行土壤有机质预测可得到精度较高的预测结果,在土壤有机质的预测与制图中具有一定的应用潜力。     

Double-sampling to improve signal-to-noise ratio (SNR) of dynamic spectrum (DS) in full spectral range

To balance the signal-to-noise ratio (SNR) of dynamic spectrum (DS) in the full spectral range due to the light intensity and the tissue absorption, we present a double-sampling method which is implemented by two spectral samplings at different integration times. In the first spectral sampling, we obtain a part of DS in the spectral range with high SNR. For the second, we increase the integration time to enhance the sensitivity of DS in original low SNR range. The combinative spectrum of two samplings is defined as the final DS for the continued model input variables. The experimental results demonstrated that the double-sampling method can widen the effective spectral range and significantly improve the SNR of DS.     

Truncated sinc slice excitation for 31P spectroscopic imaging

The shortest possible delay (Td) between slice selection and data acquisition is important for producing high quality 31P spectra. In single slice multivoxel spectroscopic imaging, conventional excitation using sinc-shaped rf pulses within typical gradient limitations can have values of Td that lead to significant spectral distortion and loss of signal. Truncated sinc excitation, which ends the excitation close to the center of the main rf lobe has been suggested for MR angiographic applications to produce short values of Td. In this work, the slice profiles, spectral signal-to-noise ratio (SNR) and spectral distortions are compared using the minimum delay achievable on a commercial MRI system for conventional 'sinc' rf excitation and truncated sinc excitation. Slice profiles are calculated using the Bloch equations and measured with a phantom. SNR and spectral distortions are evaluated from whole slice spectra on a human volunteer. On an MRI system with 1 G/cm gradients (0.5 msec risetime), for a 2.5-cm slice at 31P frequencies, conventional excitation can be adjusted to achieve Td = 2.5 msec while truncated sinc excitation yields Td = 1.5 msec. The truncated sinc excitation's shorter value of Td leads to much smaller spectral distortions, but its slice profile has "dispersive tails" which increase as more of the rf is truncated. Slice profile corrected SNR for the beta-ATP peak of 31P on a human volunteer is equivalent for both sequences while, qualitatively, in the PDE region the truncated sinc approach has improved SNR.     

Experimental relative line strengths within selected multiplets of neutral sulfur from the visible and infrared spectral range

In a wall-stabilized cascade arc helium plasma with some traces of SF₆ has been generated. On the basis of measured total line intensities emitted from the plasma, the relative line strengths within twelve multiplets of neutral sulfur (SI) have been determined. The studied lines cover the spectral range from 415 nm to 1046 nm. The obtained experimental line strengths are compared with results of recent theoretical data as well as with results of older experiments. In addition the SI line strength data are compared with analogous transitions in the OI spectrum.     

Laser-Induced Breakdown Spectroscopy: An Approach to Detect Adulteration in Turmeric

In the present paper, four turmeric samples, A, B, C, and D, were analyzed using the laser-induced breakdown spectroscopic (LIBS) technique. Laser-induced breakdown spectra of A, B, and C give the spectral lines of elements such as carbon, sodium, potassium, magnesium, calcium, iron, and molecular bands of CN, whereas spectral signatures of the toxic elements lead and chromium are also present in sample D, demonstrating the adulteration in it. Principal component analysis of the LIBS data of the turmeric samples has been used for instant discrimination. The plots and principal components provide clusters belonging to two groups, enabling laser-induced breakdown spectroscopic data along with principal component analysis to be used as an instant monitoring tool.     

Improvement of signal-to-noise ratio by stochastic resonance in sigmoid function threshold systems, demonstrated using a CMOS inverter

Stochastic resonance (SR) has become a well-known phenomenon that can enhance weak periodic signals with the help of noise. SR is an interesting phenomenon when applied to signal processing. Although it has been proven that SR does not always improve the signal-to-noise ratio (SNR), in a strongly nonlinear system such as simple threshold system, SR does in fact improve SNR for noisy pulsed signals at appropriate noise strength. However, even in such cases, when noise is weak, the SNR is degraded. Since the noise strength cannot be known in advance, it is difficult to apply SR to real signal processing. In this paper, we focused on the shape of the threshold at which SR did not degrade the SNR when noise was weak. To achieve output change when noise was weak, we numerically analyzed a sigmoid function threshold system. When the slope around the threshold was appropriate, SNR did not degrade when noise was weak and instead was improved at suitable noise strength. We also demonstrated SNR improvement for noisy pulsed voltages using a CMOS inverter, a very common threshold device. The input-output property of a CMOS inverter resembles the sigmoid function. By inputting the noisy signal voltage to a CMOS inverter, we measured the input and output voltages and analyzed the SNRs. The results showed that SNR was effectively improved over a wide range of noise strengths.     

Improved signal-to-noise ratio in spectral-domain compared with time-domain optical coherence tomography

A signal-to-noise ratio (SNR) analysis is presented for optical coherence tomography (OCT) signals in which time-domain performance is compared with that of the spectral domain. A significant SNR gain of several hundredfold is found for acquisition in the spectral domain. The SNR benefit is demonstrated experimentally in a hybrid time-domain-spectral-domain OCT system.     

Effects of temporal and spectral factors of maskers on speech intelligibility

This study demonstrates a new possibility of estimating intelligibility of speech in informational maskers. The temporal and spectral properties of sound maskers are investigated to achieve acoustic privacy in public spaces. Speech intelligibility (SI) tests were conducted using Japanese sentences in daily use for energy (white noise) or informational (reversed speech) maskers. We found that the masking effects including informational masking on SI might not be estimated by analyzing the narrow-band temporal envelopes, which is a common way of predicting SI under noisy conditions. The masking effects might instead be visualized by spectral auto-correlation analysis on a frame-by-frame basis, for the series of dominant-spectral peaks of the masked target in the frequency domain. Consequently, we found that dissimilarity in frame-based spectral-auto-correlation sequences between the original and masked targets was the key to evaluating maskers including informational masking effects on SI.     

Effects of spectral smearing and temporal fine-structure distortion on the fluctuating-masker benefit for speech at a fixed signal-to-noise ratio

Normal-hearing listeners receive less benefit from momentary dips in the level of a fluctuating masker for speech processed to degrade spectral detail or temporal fine structure (TFS) than for unprocessed speech. This has been interpreted as evidence that the magnitude of the fluctuating-masker benefit (FMB) reflects the ability to resolve spectral detail and TFS. However, the FMB for degraded speech is typically measured at a higher signal-to-noise ratio (SNR) to yield performance similar to normal speech for the baseline (stationary-noise) condition. Because the FMB decreases with increasing SNR, this SNR difference might account for the reduction in FMB for degraded speech. In this study, the FMB for unprocessed and processed (TFS-removed or spectrally smeared) speech was measured in a paradigm that adjusts word-set size, rather than SNR, to equate stationary-noise performance across processing conditions. Compared at the same SNR and percent-correct level (but with different set sizes), processed and unprocessed stimuli yielded a similar FMB for four different fluctuating maskers (speech-modulated noise, one opposite-gender interfering talker, two same-gender interfering talkers, and 16-Hz interrupted noise). These results suggest that, for these maskers, spectral or TFS distortions do not directly impair the ability to benefit from momentary dips in masker level.     

基于主成分分析的太阳光谱信息提取

太阳光谱观测是研究太阳大气活动现象有效手段之一。提出了一种基于主成分分析的太阳光谱特征信息提取和重构方法,分析了重构数据噪声抑制程度和主成分阶数的关系,计算了不同主成分阶数下重构数据的谱线信噪比以及多普勒速度测量精度。结果显示特征信息提取后,重构数据较大程度保留了原始光谱数据信息,光谱数据信噪比明显提高,谱线多普勒速度测量精度也显著提高,并且三维光谱数据存储和传输量大幅缩减。该方法能够满足一米新真空太阳望远镜当前数据规范发布需求和科学目标要求,为中国在建的光纤阵列太阳望远镜以及未来的巨型太阳望远镜光谱数据处理提供参考。