基于彩色扫描仪的图像光谱重构

针对彩色扫描仪的特点,采用主元分析法(PCA)和反向传播(BP)人工神经网络(ANN)相结合的方法对图像光谱重构进行研究。选择IT8.7/2标准色卡作为训练样本,将该色卡中的另一组色靶作为检验样本以讨论不同网络结构以及不同主元数和训练样本数对光谱重构的影响,再以自然色系统(NCS)色卡为检验样本来分析不同种类的训练和检验样本与光谱重构性能的关系。实验结果表明,采用3-14-6网络结构和6个主元数是最佳选择,训练样本和扫描目标之间的一致性是基于彩色扫描仪图像光谱重构的关键所在。     

孟塞尔色序系统与CIE1931标准色度系统转换新算法

孟塞尔色序系统到CIE标准色度系统的自动转换在颜色和视觉领域是一个很重要的课题.提出了一种基于等色调线和等彩度线插值的新算法,实现孟塞尔色序系统与CIE1931标准色度系统之间颜色的相互转换.实验测试表明,由孟塞尔系统到CIE色度系统的正向转换准确度为ΔY=0.004,Δx=0.000 13,Δy=0.000 13,由CIE色度系统到孟塞尔系统的反向转换准确度达到了ΔV ≈0,ΔH=0.017 5,ΔC=0.013.     

汽配件颜色喷涂顺序问题的TSP转化与建模

汽配件颜色喷涂顺序问题通常以生产线上相邻汽配件颜色切换次数少为最优目标,以进一步降低生产成本.该类问题具有所有汽配件都必须喷涂一次且只喷涂一次的特点,为此提出了TSP转化与建模的方法.将待喷涂汽配件定义为TSP顶点,任意两个待喷涂汽配件的颜色切换定义为顶点的距离,仿照TSP问题构建0-1规划模型;类似于顶点距离,将某些...     

地面目标红外特性及其隐身技术研究

随着红外搜索与制导技术的飞速发展,对于地面目标红外特征及其隐身技术的研究具有越来越重要的军事意义.在简化目标传热模型的基础之上,计算出目标的瞬时温度变化,并与背景的温度比较发现,目标与背景在一天之中均存在明显温差,最大温差可达10℃.从目标与背景的辐射对比度出发,分析了辐射对比度与目标表面发射率以及目标与背景温差的关系,发现了当目标与背景温差控制在±4℃之内时,采用固定发射率涂料可以使目标达到红外隐身的目的.     

Conversion from n bands color space to HSI n color space

This paper presents the method to get the equations that transform a color space of n independent primary colors to the HIS _n color system (H: hue, S: saturation, I: intensity); n indicates the number of bands and the shape of the HIS _n space. For n = 3 the structure is a double triangular pyramid, for n 4 it is the structure of the double pyramid tetrangular, and so on.     

Readily Accessible and Predictable Naphthalene‐Based Two‐Photon Fluorophore with Full Visible‐Color Coverage

Herein we report 22 acedan‐derived, two‐photon fluorophores with synthetic feasibility and full coverage of visible wavelength emission. The emission wavelengths were predicted by computational analysis, which enabled us to visualize multicolor images by two‐photon excitation with single wavelength, and to design a turn‐on, two‐photon fluorescence sensor for endogenous H₂O₂ in Raw 264.7 macrophage and rat brain hippocampus ex vivo.     

Contours of constant Δ and Ψ in the 2–4 plane for the ellipsometric functions

Contour graphs of ₂ vs ₄ for different film thicknesses and a range of angles of incidence have been plotted for the ellipsometric functions Δ and Ψ in both the reflection and transmission modes. In the case of reflection ellipsometry, when the plots for Δ_R and Ψ_R are superimposed, the two sets of contours cross nearly at right angles over a large part of the field, this being indicative of the high accuracy obtainable in using this technique to determine ₄ and ₂ and hence the optical constants, n and k, for the film material. The reflection ellipsometric technique is accurate over angles of incidence between 30° and 75° and for a range of film thicknesses between λ/30 and 5λ. Transmission ellipsometry is less useful, due to anomalies in both X_s and X_p where sudden phase changes of ±π occur in regions of interest. There is also the possibility of multiple solutions, although the use of a multiangle technique would enable the “correct” values to be more easily determined.     

归一化偏差本体反射特征的研究

使图像获得不受光照和几何条件影响的颜色特性描述是实现颜色恒常性的关系，利用双色反射模型和有限维线性模型，提取归一化偏差体反射特征。被实验证明是实现颜色恒常性的一个重要途径，本文详细阐述了归一化偏差本体反射特征的特性，其计算方法，应用的前提条件，以及应用限制，并在实际图像处理中，考虑到了这些影响的因素。     

人工神经网络法用于高效毛细管电泳分离条件优化的研究

把人工神经网络（ＡＮＮ）法应用于高效毛细管电泳（ＨＰＣＥ）分离条件的优化，给出了反向传播（ＢＰ）的ＡＮＮ模型的具体算法。用正交试验法同时考察了缓冲溶液组成、浓度、ｐＨ值和有机添加剂浓度等实验因素对ＨＰＣＥ分离合成色素和防腐剂的影响，采用误差反向传播方法建立了有效的ＡＮＮ预测模型，预测最佳分离条件，获得了满意的分离结果。