识别混凝土板内部缺陷的人工神经网络算法

以一维有缺陷混凝土板为研究对象，分别采用Leverberg-Marcluardt和径向基神经网络算法，对缺陷的深度与厚度进行识别，从而实现对混凝土板内部缺陷的三维重构，称为红外CT模拟．两类神经网络算法的识别结果表明：Leverberg-Marcluardt神经网络较径向基神经网络具有更好的收敛精度与计算效率．     

认知语言学对隐喻的诠释

世界在变,人对世界的认知也是在不断深化；对语言研究从规范、描写发展到解释认知语言学在客观主义和主观主义之间架起了一座桥梁,提出了主客观相结合的体验主义认识论。隐喻是一种重要的认知模式,是新的语言意义产生的根源。本文通过认知语言学理论阐述研究隐喻的理论基础,并解释隐喻与认知、隐喻与本义的关系,说明对隐喻的研究应注意解释的范围与意义,应从文化与思维入手,探讨语言的哲学实质,从而对语言进行深层次的研究。     

舰船·天安门·顶戴花翎——“‘洋务运动‘博物馆“东立面形象反思

通过对“洋务运动博物馆”东立面形象的反思 ,讨论了建筑形态设计中隐喻手法的运用 ;就此引申 ,探讨传统问题 ,指出正确理解传统与创新的四个方面。     

复杂背景下低信噪比点目标的实时检测算法及实现

为检测复杂背景中的低信噪比 (SNR)点目标 ,提出了一种局部自适应门限检测算法。该算法以低通滤波算法估计背景中的低频成分 ,及图像局部方差估计背景起伏 ,来计算出局部自适应的目标检测判决门限。该算法充分考虑了可实现性和实时性并采用现场可编程门阵列 (FPGA)设计实现该算法。理论分析和试验结果都表明 ,该算法可有效抑制云层、树木、地物等背景杂波、检测出复杂背景下的低信噪比点目标 ,具有恒虚警率和易于工程实现的优点。     

热轧圆钢外径的激光在线检测研讨

本文报导热轧圆钢激光在线检测外径的阶段性研究结果，给出仪器研制原理及光机结构设计同仪器精度关系的研讨．     

一种基于数字水印的错误检测和定位技术

为了增强视频通信系统在有误码信道的适应能力 ,提出了一种基于数字水印技术的错误检测方法 ,使得解码器可以对误码块进行有效的补偿。其原理为编码器在量化后离散余弦变换 (DCT)系数中嵌入易碎的数字水印 ,解码器可以通过提取水印从而确认错误以及错误发生的首宏块位置。仿真结果证明 :在信道误码率为 5× 10 - 4时 ,与常用的基于语法的方法相比 ,本方法的错误检测率是其的 1.32～ 1.82倍 ,错误正确定位率是其的 2 .95～ 5 .32倍 ,而且 PSNR只有微弱降低 ,同时算法复杂度很低     

The detection of structural deformation errors in attitude determination

In the determination of the attitude parameters from a multi-antenna GPS array, one of the major assumptions is that the body frame is rigid at all times. If this assumption is not true then the derived attitude parameters will be in error. It is well known that in airborne platforms the wings often experience some displacement during flight, especially during periods of initializing maneouvres, such as taking off, landing,and banking. Often it is at these points in time that it is most critical to have the most precise attitude parameters.There are a number of techniques available for the detection of modeling errors.The CUSUM algorithm has successfully been implemented in the past to detect small persistent changes. In this paper the authors investigate different methods of generating the residuals, to be tested by the CUSUM algorithm, in an effort to determine which technique is best suited for the detection of structural deformation of an airborne platform. The methods investigated include monitoring the mean of the residuals generated from the difference between the known body frame coordinates, and those calculated from the derived attitude parameters. The generated residuals are then passed to a CUSUM algorithm to detect any small persistent changes. An alternative method involves transforming the generated residuals into the frequency domain through the use of the Fast Fourier Transform. The CUSUM algorithm is then used to detect any frequency changes. The final technique investigated involves transforming the generated residuals using the Haar wavelet. The wavelet coefficients are then monitored by the CUSUM algorithm in order to detect any significant change to the rigidity of the body frame.Detecting structural deformation, and quantifying the degree of deformation, during flight will ensure that these effects can be removed from the system, thus ensuring the most precise and reliable attitude parameter solutions. This paper, through a series ofsimulations, will assess the effectiveness of the above mentioned techniques for detecting structural deformation effects on a GPS multi-antenna array. These principles are then tested with experimental data.     

小波多尺度分解及其在SAR图象目标检测中的应用

合成孔径雷达图象的目标检测是最重要的任务之一.目前用于目标检测的现有方法速度较慢、准确率较低、精度较差.作者提出小波分解的方法用于目标检测.同尺度目标检测采用二维离散二进小波分解方法,提高检测速度.不同尺度的目标检测采用多尺度二维离散小波分解方法,提高检测准确率.图象匹配采用灰度归一化的灰度相关匹配方法,提高匹配精度.实验证明,采用该法进行目标检测能获得较好的效果.     

基于数学形态学多级平均的图像边缘检测

提出了一种基于多级平均的数学形态学图像边缘检测方法，分析了图像边缘提取的特点．与其它形态学边缘检测方法进行了比较知，该方法在保持良好边缘的同时，提高了抗噪能力．     

交际过程中概念流利缺乏和多样性缺乏的隐喻性分析

二语学习者在交际过程中体现的概念流利的缺乏和多样性的缺乏分别由Danesi和文秋芳提出,对两者的解释都超出了语言范畴本身.本文基于概念隐喻和语法隐喻相关理论对这两种现象进行分析,认为当代隐喻理论的发展为这两种现象的解决提供了可能性,而且多样性的缺乏是概念流利缺乏的表现形式之一.