再现图像细节抑制散斑噪声技术研究

为了消除数字全息再现像产生的散斑噪声,一种基于边缘检测各向异性扩散方程的去噪算法被提出。在各向异性扩散方程(PM)基础上,通过将边缘检测算子结合梯度算子引导扩散过程,边缘检测算子对噪声具有低敏感度,可以更好地区分边缘信息,达到数字全息再现像去噪目的。通过去噪效果图对比和与中值滤波、均值滤波、小波变换滤波算法峰值信噪比(PNSR)的比较,表明本文提出的算法具有更好的去噪效果,在去除噪声的同时有效保护了图像的细节信息。     

一种基于块模糊增强的玉米田遥感图像边缘检测方法

针对高分一号卫星(GF-1)玉米田遥感图像中玉米田光谱复杂和地块边缘模糊导致的面积统计误差大的问题,本文提出一种块模糊增强和最小值边缘提取相结合的边缘检测方法进行玉米田地块分割处理,以减小面积统计误差。首先将彩色遥感图像从RGB变换到I¹I²I³彩色空间,提取出含丰富特征的单色图I¹;然后利用模糊理论对I¹进行基于块的增强处理;再对增强后的图像进行最小值边缘提取;最后利用Full Lambda-Schedule算法对区域边缘进行优化。通过与Canny和Sobel等边缘提取方法比较,证明本文的边缘检测结果能有效地分割出玉米田地块目标,减少了玉米田光谱复杂和边缘模糊带来的影响,检测出的边缘更符合玉米田实际分布,玉米田面积统计结果更符合实际。     

基于人工蜂群优化的MR图像分割算法研究

针对传统图像阈值分割算法在MR图像分割时存在的易受采集图像灰度不均、医学图像易受噪声干扰,因而难以得到准确分割阈值的问题,本文将人工蜂群算法与二维OSTU阈值分割算法相结合,提出一种基于人工蜂群优化的MR图像分割算法。使用医学图像的离散度矩阵的迹作为人工蜂群优化的目标函数,得到二维OSTU的最佳分割阈值;根据得到的最佳阈值,对图像采用二维OSTU分割的方法进行分割。实验结果证明,对于医学MR图像,本文所提出的算法具有精度高和鲁棒性强的特点,能够得到精确的分割后图像。     

一种基于概率Hough变换的遥感图像中线目标检测新方法

线状目标检测是遥感图像中人造目标识别的重要环节,其技术关键是如何有效地从背景中提取目标的长直边缘。Hough变换是从图像中检测直线的经典方法,但遥感影像的画幅大、背景复杂,标准的Hough变换会检测出大量短直线而影响对有用目标的检测效果。本文提出了一种改进的概率Hough变换方法,首先对原始遥感影像分块,利用Canny算子检测出每个子图像的边缘,再通过Hough变换检测出每个子图中的直线,并对其中的短直线聚类分组以排除干扰,最后再从剩下线段中提取并筛选出平行线以完成长直线目标检测。实验表明,这种方法对复杂背景下的遥感影像中的线状目标有良好的检测效果。     

基于分通道自适应匹配的梯度保持色彩迁移算法

本文针对经典Welsh与Reinhard算法进行改进，提出一种基于分通道自适应匹配的梯度保持色彩迁移算法。引入L、a、b三通道作为匹配依据，因通道间的序列不相关，对单一通道进行处理时不会影响到另外两通道，消除了RGB颜色模式中单一目标区域易受多通道影响的弊端；对整体灰度值相近的图像，将色度均值差异最大的待处理区与背景区通道及参考图像对应通道进行自适应匹配，解决了Welsh算法对灰度差异小或灰度范围重叠的彩色源图像分级易造成误判的问题。然后引入梯度因子，同参考图像与源图像标准差比值作加权运算，将结果作为本文改进迁移算法的缩放比例系数，对匹配到的每组像素采用保持梯度的色彩迁移算法进行着色，避免了Reinhard算法易出现色彩细节信息丢失、阶调间过渡不自然等问题。实验证明，本文改进算法所得目标图像的彩色化效果均优于经典Welsh与Reinhard算法，极大提升了色彩迁移算法的灵活性与适用性。     

数字图像处理技术在扫描电化学显微镜中的应用

扫描电化学显微镜(SECM)使用超微电极作为探针,由于发生在探针电极上的氧化还原反应是一个扩散过程,探针电极在快速移动过程中会对扩散过程产生影响,导致SECM图像变得不清晰。本研究采用Lo G算法与NEDI插值算法相结合的图像处理技术对获得的SECM图像进行处理,Lo G算法可以提高SECM图像的清晰度,但会导致图像中部分边缘信息丢失,利用基于边缘导向插值的NEDI算法对后续图像进行处理,可以很好地解决这个问题。采用离子溅射方法制备了金叉指电极和金点阵电极两种基底,对金叉指电极基底、金点阵电极基底和印有指纹的ITO基底进行了SECM成像,通过对3种基底的SECM原始图像、Lo G变换后的图像和NEDI插值后的图像进行比较分析,表明Lo G算法与NEDI插值算法结合在一起的图像处理技术可以明显提高SECM图像的清晰度和分辨率。     

基于导数直方图和神经网络的色谱基线提取算法

根据基线信号的变化特征,提出了一种新的色谱基线提取算法－基于一阶导数直方图和神经网络的色谱基线提取算法。鉴于信号总是存在着噪声,含噪声的基线信号一阶导数值也较大,并且分布无规律,而采用了多分辨率小波变换滤波技术。通过选择阈值门限,本算法可精确提取出信号的基线,尤其适用于非基线对称类的色谱信号的基线（趋势函数）提取。     

波长型SPR检测仪的灵敏度探讨

利用自行设计构建的可变入射角的波长型表面等离子体共振(SPR)检测仪, 在不同浓度的蔗糖溶液中测定了不同入射角度(80°~66°)的共振曲线, 经过处理得到共振峰位、半高宽及灵敏度随入射角和样品折射率变化的三维图像. 在此基础上探讨了波长型SPR检测仪的主要参数对仪器性能的影响, 从理论和实验上证明了影响灵敏度的主要因素为共振波长, 并且随着共振波长的增大, 检测灵敏度迅速提高.     

Au(111)面甲烷硫醇吸附的密度泛函理论研究

采用基于密度泛函理论的第一性原理方法和平板模型研究了CH₃SH分子在Au(111)表面的吸附构型和电子结构. 系统地计算了S原子在不同位置以不同方式吸附的系列构型, 计算结果表明, CH₃SH分子倾向于吸附在top位上, S-C键相对于Au表面法线的夹角为62°～78°|而S-H键断裂后CH₃S_H则倾向于吸附在bri-fcc位上, S-C键相对于Au(111)表面法线的夹角为49°～57°. 比较分析CH₃SH分子和CH₃S_H的吸附, 发现CH₃SH分子倾向于不解离吸附, 表面温度的提升和缺陷的出现可能促使S-H键的断裂. 通过比较S原子在独立的CH₃SH分子和吸附状态下的局域态密度, 发现S-H键断裂后S原子和表面的键合强于S-H键未断裂时S原子和表面的键合. 扫描隧道显微镜(STM)图像模拟显示了CH₃SH和CH₃S_H在Au(111)表面吸附的3个典型的STM图像.     

疲劳驾驶检测中基于稀疏表示的眼睛状态识别研究

为了解决传统接触式疲劳驾驶检测方法影响驾驶、检测算法识别率较低等问题,本文提出一种基于稀疏表示的眼睛状态识别的方法。利用K-SVD(K均值奇异值分解)方法对输入的训练集构造过完备冗余字典,利用正交匹配追踪法对测试的图像进行稀疏表示,然后根据重构图像和测试图像之间的误差,确定测试图像所属的类别,判断出测试图像的状态。实验中将K-SVD和OMP(正交匹配追踪)方法与其它字典学习和稀疏表示方法进行对比,结果表明,利用K-SVD字典学习算法结合OMP算法获得了较好的识别效果。     

Adaptative finite element simulation of currents at microelectrodes to a guaranteed accuracy. Application to a simple model problem

In this series of papers we consider the general problem of numerical simulation of the currents at microelectrodes using an adaptive finite element approach. Microelectrodes typically consist of an electrode embedded (or recessed) in an insulating material. For all such electrodes, numerical simulation is made difficult by the presence of a boundary singularity at the electrode edge (where the electrode meets the insulator), manifested by the large increase in the current density at this point, often referred to as the ‘edge-effect’. Our approach to overcoming this problem involves the derivation of an a posteriori bound on the error in the numerical approximation for the current that can be used to drive an adaptive mesh-generation algorithm. This allows us to calculate the current to within a prescribed tolerance. We begin by demonstrating the power of the method for a simple model problem — an E reaction mechanism at a microdisc electrode — for which the analytical solution is known. In this paper we give the background to the problem, and show how an a posteriori error bound can be used to drive an adaptive mesh-generation algorithm. We then use the algorithm to solve our model problem and obtain very accurate results on comparatively coarse meshes in minimal computing time. We give the technical details of the background theory and the derivation of the error bound in the accompanying paper.     

Faraday Ripples, Parametric Resonance, and the Marangoni Effect

In general, the combined actions of two destabilizing mechanisms do not simply add to each other. Here we show that there is a subtle interplay between parametric excitation and thermal gradients leading to interfacial instability, overstability, and generation of surface waves. The case studied refers to the stability of a liquid layer with an open free surface subjected to a transverse temperature gradient (with the Marangoni effect) and also subjected to the simultaneous action of periodic vibrations normal to the layer. Stability is examined in the weak viscosity approximation by applying a multiscale method. To a first approximation, whatever the imposed thermal gradient, vibrations with fairly large amplitude are responsible for excitation of ripples with half the imposed vibration frequency, but their amplitude depends on the Marangoni number. However, as the Marangoni number increases, the critical amplitude decreases from the excitation threshold of Faraday ripples, and after passing through a minimum it monotonically increases with increasing thermal gradient. Another salient finding is that the threshold of the Marangoni overstability is found to be independent of the imposed vibration frequency and amplitude. Copyright 2001 Academic Press.     

Application of Deep Learning Workflow for Autonomous Grain Size Analysis

Traditional grain size determination in materials characterization involves microscopy images and a laborious process requiring significant manual input and human expertise. In recent years, the development of computer vision (CV) has provided an alternative approach to microstructural characterization with preliminary implementations greatly simplifying the grain size determination process. Here, an end-to-end workflow to measure grain size in microscopy images without any manual input is presented. Following the ASTM standards for grain size determination, results from the line intercept (Heyn’s method) and planimetric (Saltykov’s method) approaches are used as the baseline. A pre-trained holistically nested edge detection (HED) model is used for CV-based edge detection, and the results are further compared to the classic Canny edge detection method. Post-processing was performed using open-source image processing packages to extract the grain size. In optical microscope images, the pre-trained HED model achieves much higher accuracy than the Canny edge detection method while reducing the image processing time by one to two orders of magnitude compared to traditional methods. The effects of morphological operations on the predicted grain size accuracy are also explored. Overall, the proposed end-to-end convolutional neural network (CNN)-based workflow can significantly reduce the processing time while maintaining the same accuracy as the traditional manual method.     

Direct one-index transformations in multiconfiguration response calculations

The evaluation of gradient vectors of a general operator with one-index, double-one-index, or higher-order one-index transformed integrals is a key operation in multiconfiguration response theory calculations. We describe an integral-driven direct algorithm that carries out this operation efficiently without pretransforming the integrals. The use of this algorithm leads to a considerable reduction in disk space and timing. © 1994 by John Wiley & Sons, Inc.     

Improvement of interfacial tension measurement using a captive bubble in conjunction with axisymmetric drop shape analysis (ADSA)

Axisymmetric drop shape analysis (ADSA) is a method to measure surface tension using drop or bubble profiles. Combining ADSA with a captive bubble configuration (ADSA-CB) facilitates pulmonary surfactant related studies. The accuracy of ADSA-CB is crucially dependent on the quality of the bubble profile extracted from the raw image. In the previous version of ADSA-CB, a global thresholding method was used to segment the bubble profile. However, that technique is of limited accuracy for images with noise and/or lack of contrast. In this paper, a new generation of ADSA-CB using the Canny edge detector was developed. To obtain better results, a novel edge smoothing technique, termed axisymmetric liquid fluid interfaces-smoothing (ALFI-S), was introduced and incorporated with the Canny edge detector to extract bubble profiles. The performance of the new version of ADSA-CB was evaluated using captive bubble images under different conditions. The results suggest that the new methodology is capable of producing accurate surface tension values under a variety of circumstances.     

TEMPERATURE-SENSITIVITY OF LIGHT-INDUCED PHASE SHIFTING OF THE CIRCADIAN CLOCK OF NEUROSPORA

Two new mutants of Neurospora craasa , designated hth-1 and hth-2 , have been isolated which allow clear expression of the circadian conidiation rhythm at high temperature (36°C). Both strains showed single-gene segregation and produced similar phenotypes but mapped to different genetic loci. These mutants allowed an analysis of the effect of temperature on (1) light-induced phase-shifting of the circadian rhythm, (2) period length of rhythm, and (3) growth rate. The amplitude of the phase response curve to light was drastically reduced as the temperature was increased from 25°C to 34°C. Phase advances were decreased more than phase delays. As previously reported (Sargent et al. , 1966), the period length of the rhythm is temperature-compensated below 30°C ( Q ₁₀˜ 1) but not well-compensated above 30°C ( Q ₁₀ 1.3–1.7). The decrease in amplitude of the light phase response curve occurred in both temperature ranges. Furthermore, the Q ₁₀ value was lowered by addition of yeast extract in the high temperature range but not in the low range. Q ₁₀ values for growth rate also differed in these strains both in the low temperature range (25–30°C) and the high temperature range (30–34°C).     

New uniform algorithm to predict reversed phase retention values under different gradient conditions

A new numerical emulation algorithm was established to calculate retention parameters in RP-HPLC with several retention times under different linear or nonlinear binary gradient elution conditions and further predict the retention time under any other binary gradient conditions. A program was written according to this algorithm and nine solutes were used to test the program. The prediction results were excellent. The maximum relative error of predicted retention time was less than 0.45%.     

Numerical calculation of overlap and kinetic integrals in prolate spheroidal coordinates. II

The efficient algorithm calculating the overlap and the kinetic integrals for the numerical atomic orbitals is presented. On the basis of the prolate spheroidal coordinates, the overlap and the kinetic integral are reduced to the integral over the rectangular domain. The integration over the rectangular domain is performed by the adaptive integration scheme. The developed algorithm is applied to calculate the integrals for the pairs of hydrogen and gallium eigenfunctions. It is demonstrated that high accuracy can be obtained for small number of integrand evaluations what guarantees the efficiency of the presented algorithm. © 2007 Wiley Periodicals, Inc. Int J Quantum Chem, 2008     

FLASH INDUCED OSCILLATION OF THE LOW TEMPERATURE THERMOLUMINESCENCE BAND Zv IN CHLOROPLASTS

Using preilluminating flashes at 2°C prior to continuous excitation of isolated chloroplasts at -80°C, a period-four oscillation with maxima at the 0th, 4th and 8th flashes was observed in the amplitude of the thermoluminescence band Z_v. Inactivation of the water-splitting system by hydroxylamine- or Tris-treatments greatly diminished the Z_v band. On the other hand, its amplitude was not considerably affected by the addition of inhibitory concentration of DCMU. On the basis of the results, the possible origin of the Z_v band is discussed.