Adaptive Multi-Feature Fusion for Vehicle Micro-Motor Noise Recognition Considering Auditory Perception

The deployment of vehicle micro-motors has witnessed an expansion owing to the progression in electrification and intelligent technologies.However,some micro-motors may exhibit design deficiencies,component wear,assembly errors,and other imperfections that may arise during the design or manufacturing phases.Conse-quently,these micro-motors might generate anomalous noises during their operation,consequently exerting a substantial adverse influence on the overall comfort of drivers and passengers.Automobile micro-motors exhibit a diverse array of structural variations,consequently leading to the manifestation of a multitude of distinctive auditory irregularities.To address the identification of diverse forms of abnormal noise,this research presents a novel approach rooted in the utilization of vibro-acoustic fusion-convolutional neural network(VAF-CNN).This method entails the deployment of distinct network branches,each serving to capture disparate features from the multi-sensor data,all the while considering the auditory perception traits inherent in the human auditory sys-tem.The intermediary layer integrates the concept of adaptive weighting of multi-sensor features,thus affording a calibration mechanism for the features hailing from multiple sensors,thereby enabling a further refinement of features within the branch network.For optimal model efficacy,a feature fusion mechanism is implemented in the concluding layer.To substantiate the efficacy of the proposed approach,this paper initially employs an augmented data methodology inspired by modified SpecAugment,applied to the dataset of abnormal noise sam-ples,encompassing scenarios both with and without in-vehicle interior noise.This serves to mitigate the issue of limited sample availability.Subsequent comparative evaluations are executed,contrasting the performance of the model founded upon single-sensor data against other feature fusion models reliant on multi-sensor data.The experimental results substantiate that the suggested methodology yields heightened recognition accuracy and greater resilience against interference.Moreover,it holds notable practical significance in the engineering domain,as it furnishes valuable support for the targeted management of noise emanating from vehicle micro-motors.

The dual-pathway model of auditory signal processing

Similar to the visual dual-pathway model, neurophysiological studies in non-human primates have suggested that the dual-pathway model is also applicable for explaining auditory cortical processing, including the ventral ＂what＂ pathway for object identification and the dorsal ＂where＂ pathway for spatial localization. This review summarizes evidence from human neuroimaging studies supporting the dual-pathway model for auditory cortical processing in humans.

Auditory hallucinations: A review of the ERC “VOICE” project

In this invited review I provide a selective overview of recent research on brain mechanisms and cognitive processes involved in auditory hallucinations. The review is focused on research carried out in the "VOICE" ERC Advanced Grant Project, funded by the European Research Council, but I also review and discuss the literature in general. Auditory hallucinations are suggested to be perceptual phenomena, with a neuronal origin in the speech perception areas in the temporal lobe. The phenomenology of auditory hallucinations is conceptualized along three domains, or dimensions; a perceptual dimension, experienced as someone speaking to the patient; a cognitive dimension, experienced as an inability to inhibit, or ignore the voices, and an emotional dimension, experienced as the "voices" having primarily a negative, or sinister, emotional tone. I will review cognitive, imaging, and neurochemistry data related to these dimensions, primarily the first two. The reviewed data are summarized in a model that sees auditory hallucinations as initiated from temporal lobe neuronal hyper-activation that draws attentional focus inward, and which is not inhibited due to frontal lobe hypo-activation. It is further suggested that this is maintained through abnormal glutamate and possibly gamma-amino-butyric-acid transmitter mediation, which could point towards new pathways for pharmacological treatment. A final section discusses new methods of acquiring quantitative data on the phenomenology and subjective experience of auditory hallucination that goes beyond standard interview questionnaires, by suggesting an iP hone/iP od app.

Speech intelligibility and auditory perception of pre-school children with Hearing Aid,cochlear implant and Typical Hearing

Purpose:There is a growing interest in speech intelligibility and audito ry perception of deaf children.The aim of the present study was to compare speech intelligibility and auditory perception of pre-school children with Hearing Aid(HA),Cochlear Implant(Cl),and Typical Hearing(TH).Methods:The research design was descriptive-analytic and comparative.The participants comprised 75 male pre-school children aged 4-6 years in the 2017-2018 from Tehran,Iran.The participants were divided into three groups,and each group consisted of 25 children.The first and second groups were respectively selected from pre-school children with HA and CI using the convenience sampling method,while the third group was selected from pre-school children with TH by random sampling method.All children completed Speech Intelligibility Rating and Catego ries of Auditory Performance Questionnaires.Results:The findings indicated that the mean scores of speech intelligibility and auditory perception of the group with TH were significantly higher than those of the other groups(P<0.0001).The mean scores of speech intelligibility in the group with CI did not significantly differ from those of the group with HA(P<0.38).Also,the mean scores of auditory perception in the group with CI were significantly higher than those of the group with HA(P<0.002).Conclusion:The results showed that auditory perception in children with CI was significantly higher than children with HA.This finding highlights the importance of cochlear implantation at a younger age and its significant impact on auditory perception in deaf children.

Auditory neuropathy in a patient with hemochromatosis

Objective:To evaluate the auditory function of an individual with genetically confirmed hemochromatosis. Methods: A 57 year old male with mildly impaired sound detection thresholds underwent a range of behavioural, electroacoustic and elec-trophysiologic assessments. These included the recording of otoacoustic emissions and auditory brainstem responses, measurement of monaural temporal resolution and evaluation of binaural speech processing. Findings for this patient were subsequently compared with those of 80 healthy controls with similar audiometric thresholds. Results: The patient showed the three cardinal features of auditory neuropathy, presenting with evidence of normal cochlear outer hair cell function, disrupted neural activity in the auditory nerve/brainstem and impaired temporal processing. His functional hearing ability (speech perception) was significantly affected and suggested a reduced capacity to use localization cues to segregate signals in the presence of back-ground noise. Conclusion:We present the first case of an individual with hemochromatosis and auditory neuropathy. The findings for this patient highlight the need for careful evaluation of auditory function in individuals with the disorder.

Auditory Plasticity Induced by Cochlear Implants in Early Life

人工电子耳蜗植入是对重度或完全感音神经性聋患者建立听觉功能的最有效装置。但是不同个体耳蜗植入后的效果,即言语识别能力之差别甚大。除了家庭、社会所能提供的术后服务与支持的差别以外,电子耳蜗移植的效果很大程度上决定于植入者中枢神经系统在电子耳蜗下发生可塑性改变的能力。另一方面电子耳蜗和其他电刺激装置的出现对听觉中枢可塑性改变和机制的研究提供了有力的工具。本文从电子耳蜗临床应用的角度出发,重点总结了近年来国外关于电子耳蜗或其他类似慢性电刺激装置植入后在先天性或语前聋个体听觉中枢通路产生的可塑性改变的研究进展,通过对已有资料的分析,探讨可塑性改变的特征、影响因素、可能机理以及对临床工作的意义。

采用生理响度感知机制的车内噪声声品质预测

现有车内噪声声品质预测的响度计算中没有考虑真实人耳生理解剖结构的传声特性,因此提出了一种基于生理响度感知模型的车内噪声声品质预估方法。首先,采集两款轿车的车内噪声样本,并通过主观评价试验得到车内噪声的主观评价烦恼度。之后,整合中耳集总参数模型与耳蜗传输线模型,构建生理响度模型。然后,以生理响度模型的响度计算值为主要参数,结合尖锐度、粗糙度与车内噪声的主观评价值,通过TabNet模型构建了车辆声品质预测模型。最终,对比分析了所构建声品质模型与基于现有标准响度模型所构建的声品质模型的预测效果。结果表明,采用生理响度模型的声品质预测平均误差百分比仅有4.73%,优于采用Moore响度(6.13%)与Zwicker响度(6.94%)的声品质预测结果。此外,所构建的TabNet声品质预测模型的平均误差百分比也低于基于BP神经网络模型的平均误差百分比(7.60%)。采用生理响度模型的TabNet声品质预测能够提高车内噪声声品质客观评价的准确率。

航迹先验融合特征的车载雷达实例分割算法

点云实例分割是场景感知中的基本任务。近年来,随着车载毫米波雷达分辨能力的提高,大量基于毫米波雷达散射点的实例分割方案被提出。实例分割的结果可作为跟踪的输入,跟踪得到各个实例的航迹信息,为后续的车辆决策与路径规划提供数据支持。然而,面向毫米波雷达的实例分割方法仍存在以下挑战。一方面,相较于激光雷达,毫米波雷达观测下的散射点更稀疏,信息量较少。当同一实例的散射点距离较远或者多个相邻实例密集分布时,分割性能显著下降;另一方面,雷达穿透性有限,路面障碍物或交通参与者对实例造成部分遮挡时,分割算法无法对实例进行正确分割和判别。考虑到实际行车场景的时间连续性,利用交通参与者的航迹先验信息,即该参与者上一时刻和当前时刻的位置信息,可以克服上述问题。因此,本文提出了一种利用航迹先验融合上一帧散射点特征的车载雷达点云分割算法。该算法利用航迹的连续性,在相邻两帧之间计算实例和散射点的对应关系并基于上述关系完成散射点特征融合。相较于单帧,融合后的高质量特征不仅信息更丰富,不同实例间的特征差异更明显,而且能弥补由于遮挡导致的信息缺失。实验结果显示,所提算法的平均覆盖率和平均精度指标分别优于基于单帧的分割算法6.19%和4.54%。该结果表明,所提算法优于文献中其他方法,能有效解决上述分割算法存在的问题。此外,与基于单帧的分割方案在典型场景的可视化对比中,所提方法也凸显了其有效性和潜力。未来,我们将进一步挖掘轨迹先验信息,以加强特征提取,同时深入探讨分割性能与帧数之间的关系。

全局感知与稀疏特征关联图像级弱监督病理图像分割

弱监督语义分割方法可以节省大量的人工标注成本,在病理全切片图像(WSI)的分析中有着广泛应用。针对弱监督多实例学习(MIL)方法在病理图像分析中存在的像素实例相互独立缺乏依赖关系,分割结果局部不一致和图像级标签监督信息不充分的问题,该文提出一种全局感知与稀疏特征关联图像级弱监督的端到端多实例学习方法(DASMob-MIL)。首先,为克服像素实例之间的独立性,使用局部感知网络提取特征以建立局部像素依赖,并级联交叉注意力模块构建全局信息感知分支(GIPB)以建立全局像素依赖关系。其次,引入像素自适应细化模块(PAR),通过多尺度邻域局部稀疏特征之间的相似性构建亲和核,解决了弱监督语义分割结果局部不一致的问题。最后,设计深度关联监督模块(DAS),通过对多阶段特征图生成的分割图进行加权融合,并使用权重因子关联损失函数以优化训练过程,以降低弱监督图像级标签监督信息不充分的影响。DASMob-MIL模型在自建的结直肠癌数据集YN-CRC和公共弱监督组织病理学图像数据集LUAD-HistoSeg-BC上与其他模型相比展示出了先进的分割性能,模型权重仅为14 MB,在YN-CRC数据集上F1 Score达到了89.5%,比先进的多层伪监督(MLPS)模型提高了3%。实验结果表明,DASMob-MIL仅使用图像级标签实现了像素级的分割,有效改善了弱监督组织病理学图像的分割性能。

基于局部感知的点云语义分割方法

点云语义分割技术是点云数据处理、三维场景理解与分析的有效手段之一。针对点云场景中局部形态各异,导致网络模型识别特征困难的问题,提出了邻域分布关系学习和混合尺度融合的方法,来增强局部感知能力。在卷积算子思想的基础上,根据邻域内所有点在三个坐标轴方向上的联合分布,学习其在高维特征层面的关系,从而捕获局部的整体相关性。此外,将包含小范围底层特征和大范围深层特征的邻域进行整体融合,有效保留不同层级的特征,并能够辅助网络修正相似或错误特征。在场景分割数据集S3DIS、ScanNet上进行实验验证,结果表明该方法在总体精度和类均精度的评价指标上均有提升,证明了其有效性。

边缘增强结合多尺度信息融合的皮肤病变分割

针对皮肤病灶大小不一、与周围皮肤对比度低、边界模糊不规则、存在伪影及毛发干扰等问题,本文提出一种边缘增强结合多尺度信息融合的皮肤病变分割算法,该算法由编码器、多尺度感知模块、边缘增强模块、轻量化解码器组成.首先在编码器中构建Transformer模块以提取全局信息,使用卷积操作以提取局部信息;其次设计多尺度感知模块,采用密集连接结构的门控空洞卷积金字塔模块来融合多尺度特征;接着构建边缘增强模块,利用深层特征促进对边缘特征的探索以更好的保留细节和边缘信息;最后设计轻量化解码器,采用CARAFE轻量化算子进行上采样,在使用相对较少参数的情况下保持较高的分割精度.在公开数据集ISIC2016和ISIC2018上做对比实验,结果表明,本文算法分割精度高于其他流行算法.

结合高频感知的大气偏振模式生成方法

大气偏振模式是一种稳定的自然属性,其在导航、探测等领域有广泛的应用,但由于自然环境以及周边建筑物遮挡的影响,在同一时刻获取的大气偏振信息是局部且不连续的,导致其在实际应用中受到影响。现有方法主要对大气偏振模式进行大范围图像的修复,对于高频信号的修复精度十分有限导致边缘模糊。针对该问题,本文采用软分割软合成的方法,对偏振信息进行冗余分割并合成,避免了高频信号的丢失,挖掘每个局部中的高频信号特征,并根据大气偏振模式时空连续性合理推测,保证重构的信息与真实信息维持一致,从而生成完整连续的大气偏振信息。实验结果证明,本方法能够很好地重构大气偏振模式中缺失的偏振信息,在云层干扰大于40%的实测重构实验中,本文方法的SSIM和PSNR得分相较于其他方法提高了26%和12%。

基于边缘引导和多尺度感知的遥感图像道路提取

为了解决遥感道路提取中边缘细节特征利用不充分,以及复杂背景遮挡区域的道路难以实现准确分割等问题,提出一种基于边缘引导和多尺度感知的遥感道路提取模型(Edge-guidance and Multi-scale perception U-Net,EMUNet)。以U-Net为基础,增加遥感图像的Canny边缘检测结果作为输入,并通过设计结合注意力的边缘引导融合模块对各层编码器进行特征引导,以此充分利用边缘信息提高最终的道路提取质量;其次,针对图像中存在的背景遮挡问题,通过构建多尺度并行空洞卷积模块增强网络的多尺度感知能力,从而捕获更多的上下文信息,对一些受到背景遮挡的区域实现准确提取。在Massachusetts道路数据集上进行实验验证,与U-Net相比,EMUNet能实现对细小道路和受遮挡区域更准确的分割,并且召回率、F1分数和交并比均优于其他对比算法,能够实现更为完整和准确的道路信息提取。

基于光纤传感的隧道管片结构形态感知试验及数值模拟研究

为精确感知环境荷载作用下隧道管片的结构形态,采用光纤传感技术,构建表贴式光纤传感元件纤芯层-防护层-基体层-被测层的多层力学模型,分析光纤应变传递机制,推导表贴式光纤传感元件的应变传递率公式,验证光纤传感元件对隧道管片结构形态感知的精度。采用1∶20缩尺模型开展试验研究,测量不同水压、土压力下管片的应变变化。结合ANSYS数值模拟,对比分析在有水压与无水压2种荷载工况下隧道管片结构的应变响应,揭示水-土耦合压力对隧道管片结构变形的影响机制。结果表明:1)在10 mm光栅长度下,光纤传感元件的平均应变传递率可达到95.1%;2)均布水压的存在对隧道管片结构变形具有一定的稳定效果;3)水-土耦合压力作用下,隧道管片结构拉、压应变变化呈现不同步现象,其中拱腰处压应变变化较拱顶处拉应变变化更为提前,表明拱腰区域对水-土耦合压力的作用更为敏感;4)有水压地层环境下,当地层荷载为0~1.6 MPa时,隧道管片结构位移呈现波动变化趋势;5)当荷载进一步达到1.6~2.4 MPa时,隧道管片结构位移呈线性增加趋势。

稳态和动态虚拟听觉重放对近场距离感知的影响

虚拟听觉重放是研究距离感知的一种重要工具,它可分为传统的稳态虚拟听觉重放和动态虚拟听觉重放。稳态虚拟听觉重放经常会产生头中定位而影响距离感知的判断,从而影响最终的实验结果。过去的实验大多采用稳态虚拟听觉重放,因此其结论也存在较大的争议性。该文对比研究稳态和动态的虚拟听觉重放对绝对距离感知效果的影响。实验采用非个性化头相关传输函数进行双耳合成,将分布在水平面0.20∼1.00 m的7个距离、0°∼180°的7个方位角下的信号通过耳机进行虚拟重放。心理声学实验结果表明,在稳态虚拟听觉重放条件下,受试者不同程度地报告了存在头中定位。而动态虚拟听觉重放系统能够利用动态因素,从而产生头外化的听觉感知。统计分析表明动态虚拟听觉重放下的距离感知效果与稳态重放存在明显的差异。由于动态虚拟重放更接近真实声源的感知效果,应选用这种方法作为实验工具。同时,在实际的空间声应用中,也应该采用动态虚拟听觉重放产生不同的距离感知。

自动驾驶环境感知多任务去耦-融合算法

自动驾驶车辆在行驶过程中,需要对行人和车辆同时完成目标检测、实例分割和目标跟踪三个任务。提出一种基于深度学习的环境感知模型同时对三个任务进行多任务学习。首先,通过卷积神经网络对连续帧图像提取时空特征;然后,通过注意力机制对时空特征进行去耦再融合,充分利用任务间的相关性,实现不同任务对时空特征的差异化选择;最后,为平衡不同任务间的学习速率,使用动态加权平均的方式对模型进行训练。在KITTI数据集上的实验结果表明,所提模型在目标检测方面,比CenterTrack模型F1得分提高了0.6个百分点;在目标跟踪方面,比TraDeS(Track to Detect and Segment)模型多目标跟踪精度(MOTA)提高了0.7个百分点;在实例分割方面,比SOLOv2(Segmenting Objects by LOcations version 2)模型AP_(50)和AP_(75)分别提高了7.4和3.9个百分点。

身材认知在服装分割设计中的感性工学应用

为探讨服装分割比例与颜色明度变化对身材认知的影响规律,结合感性工学的基本原理、方法与认知行为学的技术对消费者在服装设计中对于身材认知的感性评价进行量化分析,通过大量基础调研制成108张实验刺激图片,以年轻消费群体为主要调查对象,采集67份实验数据。研究结果表明:服装分割比例和颜色明度对于服装的显瘦感、喜好感呈现规律基本一致;最深明度的颜色显瘦感、喜好感最高,但在次深明度时的显瘦感、喜好感并未次高,所有的感知评分也未符合大众传统认知的“颜色明度越深、显瘦感越高”的规律;未出现极端化评分的样本,认知情感表征不显著。研究成果有助于提升产品的开发效果,调节消费者对身体的满意度,并为服装艺术设计、服装消费者与服装品牌提供一定的参考。

联合语义分割和深度估计的交通场景感知算法

受不同像素级视觉任务间的特征信息能够相互指导和优化的思路启发,基于多任务学习理论提出联合语义分割和深度估计的交通场景感知算法.提出双向跨任务注意力机制,实现任务间的全局相关性显式建模,引导网络充分挖掘和利用任务间互补模式信息.构建多任务Transformer,增强特定任务特征的空间全局表示,实现跨任务全局上下文关系的隐式建模,促进任务间互补模式信息的融合.设计编-解码融合上采样模块来有效融合编码器蕴含的空间细节信息,生成精细的高分辨率特定任务特征.在Cityscapes数据集上的实验结果表明,所提算法的语义分割平均交并比达到79.2%,深度估计均方根误差为4.485,针对5类典型交通参与者的距离估计平均相对误差为6.1%,能够以比现有主流算法更低的计算复杂度获得更优的综合性能.

基于光照感知和特征增强的可见光-热红外图像语义分割

在智能光电设备中,基于人工智能的可见光-热红外(Red Greed Blue-Thermal, RGB-T)图像语义分割任务可以广泛应用于自动驾驶、无人机航拍、视频监控等。图像的光照信息能在一定程度上反映场景中图像局部区域信息的可靠性,利用光照先验信息有助于进一步提高语义分割的性能。基于此,提出一种基于光照感知和特征增强的RGB-T图像语义分割模型,通过挖掘光照先验信息并结合注意力机制,引导网络在多模态图像特征融合过程中更加关注可靠信息的提取,同时抑制干扰信息的引入。实验在MFNet数据集上与最新的12种方法进行了比较,相比于性能第2的模型,mAcc提高了5.4%,mIoU提高了1.0%。所提网络模型能够获得更准确的分割结果,并通过定性定量实验验证所提模型及各个模块的有效性。

基于多尺度区域可靠性感知的半监督肺肿瘤分割

提出一种基于多尺度下一致性和区域可靠性感知的半监督学习方法,在使用少量标注数据的情况下结合未标记的数据来实现高性能的肺部肿瘤分割任务。首先,提出一种多尺度一致性均值教师框架,构建多尺度一致性损失并约束教师学生网络中多个尺度上的输出保持一致,使模型学习到更丰富的一致性知识。此外,提出一种区域可靠性感知方法使一致性学习之间的知识交换更加有效,使模型从无标注的数据中学习到更有效且可靠的知识。本文方法在医学图像分割十项全能比赛肺肿瘤数据集上进行充分的评估,与当前先进的半监督学习方法比较有更优越的性能,验证本文方法的有效性。