基于文本句法的文本生成模型
[作者]张晓龙; 姚天顺;
[摘要]本文引用语言学及符号学理论，综合自然语言文本生成所涉及的知识，构造了计算机文本生成的理论模型。文中探究了文本句法中的微观整合性和宏观整合性以及它们对文本生成的作用。这个以文本结构模型和文本意义模型为主体的文本生成理论框架，为进一步的计算机语言生成研究提供了方法依据。
[Abstract]ccording to theon'es of Linguistics and semiology,this paper analyses some knowledge of text generation in nature language,proposes a theorical model of computer text-generation. We explores microcosm-entirety and macrocosm-entirety of text-Syntax and function of them in the generation model.This text generation theorical frame based on text structure model and text sense model gives a mothod foundation for deep research of computer text generating.
[关键字]语义分析; 自然语言理解; 符号学; 文本生成;
|
基于笔划特征的宋体字形衍生方法
[作者]严伟荣; 蔡士杰;
[摘要]本文介绍了基于笔划特征的轮廓汉字字形衍生方法，并叙述了宋体轮廓汉字衍生系统在微型计算机上的设计与实现。
[Abstract]Abstract This paper introduces an approach to derive curve-outline Chinese character which is based on stroke feature,and describes the design and implementation of a Song Ti typeface deriving system on micro-computer.
[关键字]笔划特征; 黑度; 宽度因子; 模板;
|
汉语文－语转换中的语言学处理
[作者]蔡莲红; 魏华武; 周俏峰;
[摘要]本文介绍了一个计算机实现的汉语文－语转换系统，它以词为单位，将文本按句子输出，可保持自然语言的韵律。为了改善合成语音的自然度和可理解度，我们总结了一些语言学规则，并应用于该系统中，得到了很好的测试结果。
[Abstract]Abstract This article covers text-to-speech synthesis research on Chinese,aimed at improving the naturalness and intelligibility of computer synthesized speech.After collected Chinese linguistical rules, a system is carried out using syllable-length waveforms,which outputs real-time speech in words and phrases,sentence by sentence, retaining Chinese speech prosody.As one step of linguistical processing,an algorithm of word cutting is proposed,which is simple and effective.Test scores on the system arc high.
[关键字]汉语语音; 文－语转换; 分词; 韵律;
|
我国汉字识别技术的历史、现状和展望
[作者]张炘中;
[摘要]本文论述了我国印刷体汉字识别、手写印刷体汉字识别、联机手写汉字识别的历史和现状．从实用化的角度指出了汉字识别技术今后的发展动向．
[Abstract]Abstract The history and present condition for printed Chinese character recognition,Handprinted Chinese Character recognition,On-line handwritten Chinese character recognition in China is discussed.The trend of Chinese Character recognition technology in the future is proposed with practical view.
[关键字]识别系统; 印刷体汉字识别; 联机手写汉字识别;
|
异或哈希算法查找中文词组性能评价
[作者]林亚平;
[摘要]本文根据汉字机内码的特点，利用异或哈希算法建立中文词库和查找中文词组．根据不同规模的中文词库，给出了相应的改进算法。对实际的中文词库测试表明，此哈希算法分布均匀、冲突较少，且速度快，因此具有较好的实用性。
[Abstract]Abstract Based on the propertics of Chinese characters internal code, this paper uses exclusive-OR hashing algorithms to create Chinese words database and search the words.The improued algorithms for the diHirent scale Chinese words database are presented. The real tests show that the algorithms have uniform distribution, less collisions and fast operation. The algorithms arc expected to be particularly useful in the Hal application.
[关键字]哈希算法; 汉字内码; 平均查找长度; 词库; 哈希函数; 哈希表;
|
手写印刷体汉字识别中近似字的区分
[作者]胡家忠;
[摘要]本文详细地分析了手写印刷体汉字中字形相似字的特点。采用对汉字点阵图象进行方向变换的方法，利用汉字的笔道（黑点）方向和背景（白点）的封闭率特征表现近似字的局部差异，在相似字特征判定表的引导下，对字形相近的字进行判别，取得了令人满意的效果。
[Abstract]
[关键字]手写汉字识别; 形相近; 封闭率; 背景场;
|
等线体和圆头体曲线轮廓字形的自动生成系统
[作者]武港山; 叶晓璐; 蔡士杰; 陆波;
[摘要]本文叙述了等线体和圆头体汉字曲线轮廓字形的自动生成系统．该系统从黑体汉字曲线轮廓字形中自动抽取骨架单线体，结合等线体和圆头体汉字构字规则自动生成多种笔划粗细规格的等线体和回头体汉字曲线轮廓字库，具有成本低、速度快、质量好等优点。
[Abstract]Abstract This paper describes a method of automatically generating curve outlined Chinese character for the typefaces of DENXIANTI and YUANTOUTI. the method extracts the frame from HEITI typeface,and then generates the curve outline of DENXIANTI and YUANTOUTI characters in many stroke-width classes. It is characterized as a low cost,high speed, and quality method.
[关键字]单线体; 汉字字形; 曲线段; 笔划宽度; 笔划中心线; 自动生成系统; 轮廓字形; 曲线轮廓字库; 映射点;
|
精密曲线轮廓字模的生成方法研究
[作者]瞿洋; 张喜斌; 苏恩泽;
[摘要]本文介绍了一种精密曲线轮廓字模生成方法．这种方法运用图象处理技术，对字符的点阵进行部件分离、边缘发现、边缘跟踪、边缘修改，以便得到精确的字符边界轮廓．然后，运用本文设计的新的邻边加权平均样条曲线量化方法，生成高精密度、高压缩比的曲线轮廓字模．实验结果表明，本文设计的方法较好地保留了字符的特征，具有较高的压缩比和较快的还原速度．该方法还可以对其它任意二值图象进行曲线量化。
[Abstract]n this paper,a new method of obtaining precise curvilinear outline matrixes is proposed. By employing image processing techniques, this method is adopted to realise component segmentation, edge detection, edge tracing and edge modification for character Point lattices so as to abstract accurate edges of characters. Then curvilinear outline matrixes of high precision and a high compressing ratio is obtained by using a new local cubic spline interpolation curvilinearization of adjacent boundary weighted avera...
[关键字]区域填充; 边缘发现; 边缘跟踪; 矢量化; 曲线化; 局部三次样条插值; 加权平均;
|
多维参数控制法及神笔汉字发生器
[作者]何尔恭; 薛开平; 温立新; 曾锡山;
[摘要]在所有汉字生成方法中＂部件组字＂最能有效地压缩存储字形的信息量，但由于部件受到压缩时的畸变使得它难以生成高质量汉字．本文提出了＂多维参数控制法＂有效地克服了上述缺点．它已应用于部件组字型汉字库──神笔汉字发生器，可以快速生成高质量汉字．
[Abstract]
[关键字]现代计算机技术; 汉字发生器; 汉字字形; 中文信息处理; 汉字库; 控制法; 多维参数;
|
曲线轮廓汉字的网格适配技术
[作者]胡长原; 武港山; 张福炎;
[摘要]要提高曲线轮廓汉字的还原质量，需要在字形还原时采用网格适配技术．本文介绍了网格适配的基本原理，并在分析汉字字形还原失真现象的基础上，给出了我们自己设计并实现的适合于曲线轮廓汉字特点的二种网格适配方法，即动态的补象素算法和曲线轮廓汉字的ｈｉｎｔｉｎｇ技术．
[Abstract]Grid-fitting is the additional process to the rasterization algorithm of outline characters in order to get a better appearance. This paper first introduces the basic principles of grid-fitting, and then, by analysing the trait of Chinese characters, comes to the special grid-fitting methods of outline Chinese characters. This paper also presents two effective experiments, one is the so-called'pixel compensation"algorithm, the other is the hinting method.
[关键字]曲线轮廓汉字; 网格适配（grid－fitting）; 字形扫描变换; Hinting技术;
|
共95页 当前第66页 