分离通道联合卷积神经网络的自动调制识别

针对通信信号的自动调制识别需要大量特征提取的问题,提出了一种分离通道卷积神经网络自动调制识别算法。该算法通过结合深度学习中卷积神经网络(CNN),分别提取时域信号的多通道和分离通道调制特征,再利用融合特征实现不同信号的分类。仿真结果表明,相比基于CNN的算法,所提算法在高信噪比下针对两个数据集的识别率分别提升7%和18%;此外,相比于基于特征提取的传统识别算法,其高阶调制识别性能平均提升3 dB。     

如何提高高速公路收费效率和准确性的研究探讨

随着现代计算机及信息技术的飞速发展,越来越多的智能化、信息化的技术和设备被广泛应用于智能交通系统。而车型自动分类与识别技术作为智能交通领域的一个重要的分支,得到了更多的关注和研究。本文首先介绍了车型自动分类与识别技术的现实意义以及其当前的研究内容。然后重点分析了音频车辆检测分类与识别,包括音频信号建模及特征提取和车辆音频信号的特征提取及车型分类。最后指出了尚待完善的问题。     

基于三维特征的雷达信号脉内调制识别

针对低信噪比条件下雷达信号识别算法对噪声敏感的问题，提出了一种基于三维特征的雷达信号脉内调制识别算法。该方法通过提取信号的差分近似熵、调和平均分形盒维数和信息维数特征组成三维特征向量，使用遗传算法优化的BP神经网络分类器实现雷达信号的分类识别。仿真结果表明，所提取的三维特征在信噪比为-4~10 dB变化范围内具有较好的类内聚集度和类间分离度，可以实现对不同雷达信号进行识别，证实了该方法的有效性。     

一种基于视觉感知的舰船目标智能化识别方法

为有效识别视觉系统采集的可见光图像中的舰船目标，提出了基于YOLO（You Only Look Once）网络模型改进的10层的卷积神经网络（Convolutional Neural Network,CNN）用于水面舰船目标的智能识别，通过反卷积的方法可视化CNN中不同卷积层提取到的舰船目标特征。按照传统目标识别方法提取了舰船目标的四类典型人工设计特征，将所提CNN的舰船目标识别结果与YOLO网络模型及四类人工设计特征结合支持向量机用于舰船目标识别的结果进行比较。实验结果表明，与YOLO网络模型相比，综合精确率、召回率和效率3个舰船目标识别的性能指标，改进后的CNN性能更好，从而验证了所提方法的有效性。不同数据量下采用典型特征识别舰船目标与基于深度CNN识别舰船目标的识别结果比较说明了不同类型目标识别算法的优劣势，有利于推动综合性视觉感知框架的构建。     

采用航迹特征的智能机型识别

针对空战目标识别中机型自动识别比较困难的问题，提出了采用航迹特征的智能目标识别方法。利用卷积神经网络（Convolutional Neural Network，CNN）分层学习特征的能力，训练CNN算法模型自动地从航迹数据中学习有用的特征并分类。利用沿海实地采集的15个类别的飞机航迹数据，经一系列数据预处理后作为智能识别算法的训练和测试数据，在验证实验中描述了算法网络的相关配置,对比了CNN与其他分类器的识别结果。实验结果表明，CNN具有很好的识别性能。     

一种优化的卷积神经网络调制识别算法

针对非合作接收条件下信号的调制识别问题,提出了一种基于循环谱特征和深度卷积神经网络的自动调制分类算法。该算法首先利用二值化、形态学操作等技术对循环谱数据集预处理,提高网络泛化能力;然后将数据集输入到卷积神经网络模型中,经过网络的特征提取实现分类识别。在网络中添加残差块网络增大感受野,提高特征提取能力。采用Dropout、优化函数等技术优化网络结构,防止训练过拟合。仿真结果表示,与传统方法和现有的一些深度学习调制识别方法相比,该算法在低信噪比条件下有更高的准确率,具有明显的抗噪声优势,是一个有效的调制识别算法。     

基于NeiMu的雷达信号聚类分选

针对雷达信号分选中常见的聚类数目难确定、数据簇形状识别、异常数据敏感等问题,提出了基于NeiMu(Neighboring Mutually)的雷达信号聚类分选算法。该算法首先以信号脉冲为点、各点间的欧氏距离为线构建距离矩阵,然后在进行干扰数据剔除的情况下选定合适k值完成聚类矩阵的构建,最后采取遍历聚类矩阵的方法输出聚类结果,在删除无效的聚类后实现了雷达信号的聚类分选。通过仿真可知该算法在选取合适k值的情况下具有极高的正确率,证明了其有效性和可靠性。     

一种基于星座图恢复的多进制相位调制信号识别算法

在实际调制过程中,无线电波传输多径及衰落引起的符号间干扰和信号接收端的载波频偏会造成星座图难以识别。针对这一问题,提出了一种基于星座图恢复和卷积神经网络的多进制相位调制信号识别算法。首先,设定相邻采样点距离和相位角的阈值以筛除发生符号间干扰时的采样点,保留剩余的有效采样点并形成聚类组;然后,通过旋转相邻聚类组抵消载波频偏带来的影响,实现星座图的恢复;最后,利用卷积神经网络对星座图进行特征自动提取和调制识别。实验结果表明,对于实测信号,所提算法能够较好地恢复星座图并实现BPSK、QPSK和8PSK的准确识别。最终的识别准确率达到了99.9%,较星座图恢复前提高了24.2%。     

基于卷积神经网络和注意力机制的图像检索

当前先进的图像检索方法中，存在着不能很好地分辨图像中不同区域和内容的重要性的问题，导致计算资源分配不合理、检索正确率较低等一系列结果。为了解决这些问题，提出了一种基于卷积神经网络(Convolutional Neural Network,CNN)和注意力机制的图像检索方法。首先使用卷积神经网络提取特征，然后使用注意力机制处理提取的特征,可以在计算能力有限的情况下根据图像中的内容合理分配计算资源，使图像中的突出部分得到更多的关注。最后通过融合全局平均池化层处理后的CNN特征来进行图像检索。所提方法在corel1k、corel5k和corel10k三个数据集上与其他先进的图像检索方法进行了比较，结果表明该方法能够有效提高图像检索的精确率和召回率，并且在检索返回的前k张图像的数量(top-k)增加时，仍能保持良好且稳定的检索精确率。     

一维卷积神经网络用于雷达高分辨率距离像识别

针对人工提取高分辨率距离像（HRRP）优良特征比较困难的问题，研究了基于一维卷积神经网络（CNN）的HRRP识别方法。利用CNN具有分层学习特征的能力，训练CNN自动地从HRRP中学习有用的特征并分类。在仿真实验中描述了网络的相关配置，分析了不同激活函数、不同参数、不同网络结构的识别性能，对比了CNN与其他分类器的识别结果，用可视化特征图直观地说明了CNN通过卷积层能够学习到易于分辨的特征。实验结果表明CNN具有很好的识别性能。     

Universal Coverage, Pinpoint Illustration and Profound Impact HTC 2008 Intermedia Recommendation and Introduction Meeting

Holding 88 percent of train media resources across China, based on "Time Cube", on the 2008 Intermedia Recommendation and Introduction Meeting on December 19, 2007, HTC announced its high-profile strategy of tour marketing and tapping the immeasurable potentials on the train. Meanwhile, HTC officially launched its brand new product Olympic Train on the meeting, providing a fabulous platform for all the brands that are prepared to do a smart marketing on the occasion of 2008 Olympic Games.     

Market presence in China strengthened Network growth and Dual-Chinese-Hub Strategy More frequet Flyer Benefit——Air China and shanhai Airlines Join Star Alliance

In the end of 2007, Air China Limited (hereinafter "Air China") and Shanghai Airlines, formally joined Star Alliance at a ceremony held in the new Terminal Three at Beijing Capital International Airport. With the addition of the two Chinese airlines, Star Alliance now has 19 member carriers operating 17,000 daily flights to 897 destinations in 160 countries.     

Brown＇s Blossom in Britain-China

The overall strategic relationship between U.K. and China is ＂better than ever＂, China- Britain relations have maintained good momentum, with increased high-level exchanges and the reinforcement of strategic mutual trust, with more room for growth in the future.     

Challenge and Opportunity Go Hand in Hand——What Global Economic Agitation Brings to the China's Economy?

Domino effects of the global finance agitation As the origin of the subprime crisis,US financial institutions lost USS45 billion last year,accounted for over half of the total US,EU & Japan loss.Since January 15th,Chtigroup released its financial statement of the 4th quarter,showing a net deficit of USS9.83 billion,which marked the hisoric high of its quarterly deficit.     

A World of Love 2008 Music Grand Ceremony

Under the guidance of the theme slogan of the 29th Olympic Games - One World One Dream, with dream, love and dedication as its theme, the 2008 Music Grand Ceremony ＆ 1000 People Charity Concert on 100 days cotmtdown to Paralympic Gamesis officiaily launched on December 22, 2007. The 2008 Music Grand Ceremony is sponsored by China Disabled Persons＇ Federation, and co-hosted by National Paralympic Committee of China,     

Sinosteel：Extraordinary Moves in Industry Chess

Among the Fortune 500 firms,more and more images from China emerge out.On the 2008 list,Sinosteel Corporation,a landmark group of China industries quietly squeezes into the candidate line.As the largest steel service ＆ trade enterprise,its sales reached at RMB 111.24 billion,with a year-on-year increase of 83.03 percent,and the profit grew by 180.03 percent.On the Sinosteel 2008 Work Meeting,the president Huang Tianwen uttered,＂In the coming year,we will work for US$20 billion sales,and look forward to getting a position in Fortune 500 firms for ourselves.＂     

Walk Through Unusual Five Years—— An Exhibitor＇s Canton Fair Course

When the 103rd Canton Fair is approaching, an old friend of Canton Fair, Li Ying from Jiangsu prov-ince recollected his past five years＇ experiences to the grand trade meeting. Each year is special, with some unforgettable memories colored by the year.     

China''''s National Oil Reserve Center was Born

Aseries of events that occurred in October 2007, are rewriting the history of China's energy industry: November 3 an Energy Law Committee asked the public for suggestions on improvements; forty top experts came together to form the National Energy Expert Advisory Committee on November 6; and the National Oil Reserve Center (NORC)became the nerve centre of oil reserves on November 18. Yet, behind the progress of these recent activities remains the fact that international oil prices still hover around a very high price of US$ 90/barrel, and China's oil imports are close to accounting for half of its total oil reserves.     

MUYANG GROUP

With its headquarters in the historic city of Yangzhou,Jiangsu Muyang Group Co.,Ltd has since its founding in 1967 grown into a well-known group corporation whose activities cover research ＆ development,project design,manufacturing,installation and services in a multitude of industries including feed machinery and engineering,storage engineering,grain machinery and engineering,environmental protection,conveying equipment and automatic control systems.     

2007 China Harbor Ten People

2007 China Harbor Ten People elected the entrepreneurs who contributed a lot to port economy and enterprises this year trough their talent management.These ten people embody their social responsibility,professional skills,creative ability,and charming personality.Bearing full confidence in China＇s port economy,the port entrepreneurs are brave enough to explore a brand new area,so as to promote harbor economic development. At the very beginning of 2008,China Shipping Website,Dalian Maritime University,and China Port and Shipping In- stitute jointly select the 2007 China Harbor Ten People,demonstrate port entrepreneurs＇ stunning minds,disclose their secrets to success,and providing abundant experiences to others in harbor development.The top ten people are various excellence during the ups and downs harbor development.