基于梯度广义近邻图的多模医学图像配准

提出一种融合梯度信息的医学图像配准算法。提取图像中互补的尺度空间特征点,并计算特征点周围的梯度信息,以Rényi互信息作为目标函数,利用广义近邻图估计Rényi熵。实验结果表明,在受外界因素影响较大的情况下,该算法速度较快、准确率较高,具有较强的鲁棒性。     

张量积Bézier曲面的降阶

应用张量积Bézier曲面的几何性质和遗传算法,给出了Bézier曲面的降阶。与已有的算法相比,该算法具有计算简单、逼近误差直接给出,几何直观性强等优点。     

基于本体构建技术的食品安全事件数据清洗研究

高质量的数据不仅为后续知识挖掘提供了最有效的数据源而且保证了所挖掘知识的可信度和可用度。为了确保基于食品安全事件数据所挖掘知识的有效性和真实性,本文提出了一种基于本体构建技术的食品安全事件数据清洗流程,并对所清洗的数据进行了相应的评价。本文所清洗的数据是酒鬼酒塑化剂这一食品安全事件数据,在具体清洗过程中,基于斯坦福大学医学院开发的七步法,通过Protégé这一工具,构建了相应的食品安全事件清洗本体。该研究为探究领域化数据的清洗工作提供了一种可行性的操作方案,并且对于所清洗的数据增加了一定的语义知识。     

太原空气污染监测系统中空气污染知识本体设计

目前空气污染已经成为全球面对的重大问题,建设快速直观的空气污染信息系统非常必要。研究基于智能型空气污染无线传感器网络系统,着重于系统中空气污染知识本体的规划与构建。研究利用protégé软件描述空气污染知识本体的概念属性与关系,建立空气污染相关知识本体,空气污染监测系统通过知识本体对污染情况进行的解析及推论,最后产生决策支持信息。通过在太原空气污染监测系统中实际应用,结果表明,系统在本体的支持下可快速准确判断污染的情况并采取相应的措施,具有较高的实用价值。     

Experimental investigation on color variation mechanisms of structural light in Papilio maackii ménétriès butterfly wings

The body color in animals results from billions of years of their natural evolution in order to evade natural enemies, catch quarries or display themselves beauty, in-vestigation on mechanisms of structural light is an important aspect of bionics. Based on the phenomenon of Papilio maackii ménétriès’ blue scales changing into green ones immediately after dropping some alcohol aqua on the underwing sur-face and soon returning back to the original color, the relationship between micro-structure, optics characteristic of scales and changing color effect were studied using the Olympus Stereomicroscope, Scanning Electron Microscopy (SEM), Transmission Electron Microscopy (TEM) and Ultraviolet (UV)-Visible Spectropho-tometer. The color variation mechanisms of blue scales of Papilio maackii mé-nétriès in Chinese Northeast were revealed in this paper. When visible lights trav-eled through the concaver structure with multilayer reflector and the filled medium with different refractive indices, the reflected lights in definite wavelengths pro-duced interference and color at that wavelength came into being. It has important academic reference value to biomimetics design of video stealth materials.     

Hájek-Rényi-type Inequality for a Class of Random Variable Sequences and Its Applications

In this paper,we obtain the Hájek-Rényi-type inequality for a class of random variable sequences and give some applications for associated random variable sequences,strongly positive dependent stochastic sequences and martingale difference sequences which generalize and improve the results of Prakasa Rao and Soo published in Statist.Probab.Lett.,57(2002) and 78(2008).Using this result,we get the in-tegrability of supremum and the strong law of large numbers for a class of random variable sequences.     

Congruency proportion reveals asymmetric processing of irrelevant physical and numerical dimensions in the size congruity paradigm.

When the proportion of congruent trials in conflict tasks is manipulated (e.g., Stroop, Simon), the typical result is that the magnitude of the conflict effect increases as the proportion of congruent trials increases. The present experiment investigated the influence of Congruency Proportion in the context of the Size Congruity Paradigm. Congruency Proportion had a significant impact on the Numerical Judgement Task (judging which of two numbers is numerically larger), but not on the Physical Judgement Task (judging which of two numbers is printed in a larger font). These data support the inference that physical size information is processed before, and more fluently than, numerical size information. The implications of this asymmetry are discussed in terms of the relative role of semantic and physical size information in representations of magnitude, and the role they play in both of these tasks. (PsycINFO Database Record (c) 2011 APA, all rights reserved)     

一种基于半定松弛技术的TDOA-FDOA无源定位算法

在运动目标的无源定位场景下,闭式算法在低噪声情况下可以到达克拉美罗下界(CRLB),但是这些算法往往不能适应较大的测量噪声环境。针对目前闭式算法适应大噪声能力较差这一问题,该文联合到达时间差(TDOA)以及到达频率差(FDOA),提出一种基于半定松弛(SDR)技术的无源定位算法。该算法首先构建传统闭式解的伪线性方程,其次利用随机鲁棒最小二乘(SRLS)的思想以及目标参数与额外变量之间的非线性关系,将无源定位问题转化为了具有2次等式约束的最小二乘问题;随后,将半定松弛技术应用到这一问题上,约束最小二乘问题松弛为半定规划(SDP)问题,最后,借助优化工具箱可以有效地对目标参数进行求解。该文所提出的算法不需要初始值先验条件,仿真实验表明了所提算法的有效性。

     

基于多球面封装透镜的LED微型投影仪的光源设计

提高光能利用率是现阶段LED微型投影仪的技术难题,而提高光源的取光效率则是提高整机光效的前提和根本途径。对于以LED为代表的朗伯源,其光学元件的结构是制约光能利用率的最大因素。我们提出了一种基于多球面单透镜的LED光源设计方案并进行了光学仿真,将LED的一次封装和二次光学设计整合为一体,简化了光源系统结构,取光效率提高到94.9%,同时压缩了出光角度和光学扩展量。     

一种基于蝙蝠算法的新型小波红外光谱去噪方法

针对传统的小波去噪方法容易产生信号振荡和丢失特征信息等问题,提出了一种基于蝙蝠算法的新的有效的红外光谱去噪方法。该方法创新性地运用蝙蝠算法优化了小波阈值和估计因子。其基本思想是,首先在解空间中随机生成一定规模的个体,然后根据向当前最优个体学习的方法进行速度更新,从而实现位置更新;同时,由于Levy飞行搜索策略会产生较大跳跃,利用这种不均匀、随机游走的特性可以实现对整个解空间的搜索,从而避免陷入局部极值点。CO气体红外光谱去噪实验的结果表明,利用蝙蝠算法对各个分解层的阈值和估计因子进行优化后,信噪比为84.184,均方误差为0.0006。由于更有针对性地保留了光谱信号中的特征信息并剔除了无用的噪声信息,该方法可以提高后续定性和定量分析的精度。