一种基于内容的图像检索方法

论述了基于颜色特征的图像检索和基于纹理特征的图像检索,在此基础上提出了一种利用图像两种内容特征进行图像检索的方法。     

单多普勒雷达风场反演方法敏感性试验

为了比较3种单多普勒天气雷达风场反演方法(即VAP方法、UW方法、涡度-散度法)的反演效果和适用性,用这3种方法对几种典型的理想风场进行反演和敏感性对比试验,并对造成反演效果差异的原因进行了细致的分析.试验结果表明,这3种方法对均匀风场的反演效果都很好,UW方法的误差最小且计算量较小;3种方法对切变场的反演效果类似,径向速度差值较大的UW方法相似系数却稍大,切变越大时反演误差越大;对于涡旋场3种方法反演效果都不理想,若为径向气流显著而切向气流不显著的涡旋场可以用UW方法来反演.因此,相对而言,可以选用计算量小且相似系数较高的UW方法来反演二维简单风场.而在研究和实际应用中,应该进一步探讨用假设条件更接近实际情况的反演方法来更准确地反演风场.     

基于小波变换的投影光栅条纹分析

投影光栅条纹相位法通常用于三维物体形貌的测量.采用小波变换直接提取单幅光栅条纹图像的相位分布,不需要进行相位展开,即可得到物体表面轮廓.给出了小波分析应用在空间载波光栅条纹相位分析中的理论推导证明、计算机模拟以及实验验证结果,讨论了小波分析的抗噪能力,证实了该方法的可行性.     

基于Web服务架构的数字图书馆信息检索系统的设计与实现

本文通过分析图书馆信息检索的现状，利用Web服务强大的信息交换能力，设计了一个基于Web服务架构的信息检索系统．使用该系统可以最终解决图书馆长期以来资源共建共享的困难．     

不确定知识的相似度量方法及应用

基于模糊集和粗糙集理论,给出常用的几种不确定知识的相似度量方法,并应用于信息检索系统中。建立了更符合用户信息需求的模糊相关检索模型。利用粗糙集的等价关系(近似关系)表示用户查询兴趣特征,并通过不精确的分类知识方法进行知识库中用户兴趣分类。     

科技资源图像检索技术研究与实现

科技资源具有海量、复杂、异构、地理分布广的特点。为了高效查找和使用科技资源的图像,综合运用TB IR(基于文本的图像检索)和CB IR(基于内容的图像检索)共同检索科技资源图像,设计了科技资源图像检索系统,并对系统中的图像分割、特征提取和相似度度量3种关键技术进行了研究。     

一种基于智能Agent的用户个性化信息检索系统模型

针对当前信息检索工具缺乏对知识的理解和处理能力,提出了一种基于智能Agent的用户个性化信息检索系统模型,通过多个Agent的协调工作为用户提供个性化服务。描述了该模型的系统结构及各功能模块的实现方法,分析了系统模型的工作流程,部分解决了信息检索中的文档相关性计算、搜索引擎的调度算法、数据库的构建等技术问题。     

Determination of the optical thickness and effective particle radius of clouds from transmitted solar radiation measurements

A method is presented for determining the optical thickness (τc) and effective particle radius (re) of stratiform cloud layers from transmitted solar radiation measurements. A detailed study shows that the cloud optical thickness and effective particle radius of water clouds can be determined from transmission function measurements at 0.75 and 2.13 μm, provided that the scaled optical thickness τ′0.75>1 and re>5 μm. The wavelengths adopted by our study are similar to the channels of the moderate resolution imaging spectrometer (MODIS). The proposed method is invalid for optically thin clouds since transmission at 2.13 μm is less sensitive to re. The retrieval errors of τ′0.75 and re monotonically decrease with increasing τc. For clouds having τ′0.75≥2, the retrieval errors of τ′0.75 and re are below 10% and 20%, respectively. Transmissions at 0.75 and 1.65 μm can also be used to retrieve τc and re.     

Blurred Defocused Image Restoration Based on FRFT

A technique for restoring the blurred image resulted from defocusing of the lens is proposed in this paper, which is based on fractional Fourier transform （FRFT）.The FRFT, as a powerful tool for the analysis of time-varying signals, is closely connected with the optical imaging system. FRFT also can describe optical imaging process just like Fresnel diffractions, so a defocused imaging model based on FRFT is established to explain the blur phenomena of defocusing image. The defocused imaging model is greatly different from the traditional point spread function （PSF） model, and enables to uncover the blur nature of non-focus image. Then, an image restoration method using the novel model is proposed to handle the blurred defocused image. The method adopted a new iterative phase retrieval approach which can approximately estimate phase signals from intensity signals of a single defocused image by means of FRFT. Restoring image may acquire sharp image by implementing inverse FRFT on complex image signal made from the estimated phase signals and intensity signals. Experimental results demonstrate that the method is effective in restoring blurred defocused image.     

Automatic Question Answering from Web Documents

A passage retrieval strategy for web-based question answering （QA） systems is proposed in our QA system. It firstly analyzes the question based on semantic patterns to obtain its syntactic and semantic information and then form initial queries. The queries are used to retrieve documents from the World Wide Web （WWW） using the Google search engine. The queries are then rewritten to form queries for passage retrieval in order to improve the precision. The relations between keywords in the question are employed in our query rewrite method. The experimental result on the question set of the TREC-2003 passage task shows that our system performs well for factoid questions.