FSNet: 基于频率特性的烟雾图像分割网络

烟雾图像分割是对烟雾进行识别与精准定位的基础, 是火灾预警的重要手段. 针对烟雾分割时存在过分割、欠分割以及边界拟合粗糙的问题, 本文提出一种基于频率分离特性的烟雾图像分割网络. 所提出的频率分离模块将特征图中的烟雾区域分离为低频主体部分和高频边界部分, 同时基于多任务学习设计多模块权重自适应损失函数对烟雾整体、主体、边界分别监督学习, 起到细化烟雾边界和改善烟雾整体分割结果的作用; 此外, 结合可变形卷积提出改进的空洞空间金字塔池化模块以解决其信息利用率低和特征关联性差的问题. 在对比实验中, FSNet的烟雾交并比为76.55%, 比基线网络提高了4.25%. 可视化分割结果可以看出, FSNet能有效缓解过分割、欠分割, 所得烟雾边界更平滑, 烟雾图像分割的整体性能获得较大提升.     

White‐Emissive Self‐Assembled Organic Microcrystals

Organic semiconductor micro‐/nanocrystals with regular shapes have been demonstrated for many applications, such as organic field‐effect transistors, organic waveguide devices, organic solid‐state lasers, and therefore are inherently ideal building blocks for the key circuits in the next generation of miniaturized optoelectronics. In the study, blue‐emissive organic molecules of 1,4‐bis(2‐methylstyryl)benzene (o‐MSB) can assemble into rectangular microcrystals at a large scale via the room‐temperature solution‐exchange method. Because of the Förster resonance energy transfer, the energy of the absorbed photons by the host matrix organic molecules of o‐MSB can directly transfer to the dopant organic molecules of tetracene or 1,2:8,9‐dibenzopentacene (DBP), which then emit visible photons in different colors from blue to green, and to yellow. More impressively, by modulating the doping molar ratios of DBP to o‐MSB, bright white‐emissive organic microcrystals with well‐preserved rectangular morphology can be successfully achieved with a low doping ratio of 1.5%. These self‐assembled organic semiconductor microcrystals with multicolor emissions can be the white‐light sources for the integrated optical circuits at micro‐/nanoscale.     

Path integral solution for nonlinear systems under parametric Poissonian white noise input

In this paper the problem of the response evaluation in terms of probability density function of nonlinear systems under parametric Poisson white noise is addressed. Specifically, extension of the Path Integral method to this kind of systems is introduced. Such systems exhibit a jump at each impulse occurrence, whose value is obtained in closed form considering two general classes of nonlinear multiplicative functions. Relying on the obtained closed form relation liking the impulses amplitude distribution and the corresponding jump response of the system, the Path Integral method is extended to deal with systems driven by parametric Poissonian white noise. Several numerical applications are performed to show the accuracy of the results and comparison with pertinent Monte Carlo simulation data assesses the reliability of the proposed procedure.     

频闪白光干涉技术

频闪白光干涉技术（SWLI）能够对被测系统进行动态模式的重建和表征,在微机电系统（MEMS）或微光机电系统（MOEMS）等振动样品测量中具有很高的实用性。传统频闪方法不能预先检测或分析具有单个振动激励源的微结构,而频闪白光干涉技术则解决了这一问题。为了全面了解这一新开发的技术,本文介绍了理论、技术方法和该技术各种应用的实验结果及分析。该技术为MEMS、MOEMS甚至微纳尺度生物体动态行为的重建和分析提供了一种有效的方法。实验结果和分析表明所开发的技术在实际应用中具有良好的可行性和准确性。     

神经元膜电位噪声取值范围的评估

大脑神经元的活动是在复杂的生理学环境下工作的，而生理学环境中的噪声来源于多个方面．如何定量地正确评估神经系统中的噪声环境是神经信息处理的基本问题．本文通过神经能量的计算给出了噪声对神经元膜电位及对应的能量波形产生显著影响的临界值的估计范围，从而定义了神经元赖以活动的生理学意义上的噪声环境．     

Dynamical responses for a vertical vibration model of a four-roller cold rolling mill under combined stochastic and harmonic excitations

Considering that random fluctuations can affect the rolling process of cold rolling mills, leading to abnormal cold mill performance and defects in the rolled product. Therefore, it is crucial to study the influence of random fluctuations on the system of cold rolling mills. In this paper, taking a model of the vertical vibration of a four-roller cold rolling mill under harmonic excitation as a prototype class for a real system, the effects of random fluctuations on the system response are analyzed. Firstly, based on the deterministic vertical vibration model of a four-roller cold rolling mill under harmonic excitation, a stochastic vertical vibration model of a four-roller cold rolling mill with random fluctuation as Gaussian white noise is introduced. Subsequently, the vertical vibration model of a four-roller rolling mill is theoretically analyzed by the averaging method and the stochastic averaging method, respectively. The effectiveness of the proposed theoretical method is verified by numerical simulation results, and the influences of harmonic excitation and random excitation on the system response are also investigated in detail. Finally, the results show that the noise induces the occurrence of stochastic transitions and bifurcations, and the steady-state probability density function and time history diagrams are given to further explain the existence of these dynamic phenomena. The related research can provide theoretical guidance for the realization of vibration control and reliability design in the four-roller cold rolling mill system.     

一种基于图像的非均匀单散射参与介质建模方法

针对非均匀单散射参与介质建模问题,提出一种能够保持密度场空间变化细节的建模方法.使用体数据、吸收和散射系数之比分别描述参与介质空间变化的密度分布及光线在其内部的传输特性,通过构建采集图像像素值与它们之间的表达式,将建模问题转化为非线性数值优化问题,求最优解.为了解决高分辨率体数据下大量体素密度值同时求解带来的时间开销大及数值不稳定问题,提出一种符合采集图像像素值明暗分布规律的密度场体数据初始化算法和一种多分辨率体数据逐步求精优化算法;为了快速计算大量体素密度值梯度,提出一种基于GPU的多体素并行梯度计算算法.实验结果表明,提出的方法能够根据采集图像重构烟模型,并保持其细节信息.     

江南传统民居建筑色彩的成因

民居建筑是在特定文化空间与特定地理环境内产生的独特文化现象。是民族智慧与审美的结晶。民居建筑的规模、结构、装饰的形成演变,是其民族文化发展历程的缩影。建筑与民族文化之间是互为因果的关系。"黑瓦白墙"作为江南传统民居中具有代表性的古典建筑风格,尤具有独特的文化内涵。本文以此为立足点,透过文化属性与地理环境等因素来探析江南传统民居建筑色彩的成因。     

Image-based modeling of inhomogeneous single-scattering participating media

We propose a method to reconstruct inhomogeneous single-scattering participating media, which could preserve fine high-frequency details of density field.Volumetric data and the ratio of absorption coefficient to scattering coefficient are used to describe the spatial-varying density distribution and optical properties of certain participating media, and a function between the above parameters and captured pixel values is built.Thus the problem of how to find solutions to these parameters is formulated into...