权益分类	功能权益	普通用户	{{item.name}}会员
{{category.name}}	{{benefitItem.name}}

Knowledge Representation about Images in a Natural Language and its Use.

自然语言图像的知识表示及其使用。

基本信息

批准号：
01580019
负责人：
NAGAHASHI Hiroshi
金额：
$ 1.02万
依托单位：
Tokyo Institute of Technology (1990)Yamagata University (1989)
依托单位国家：
日本
项目类别：
Grant-in-Aid for General Scientific Research (C)
财政年份：
1989
资助国家：
日本
起止时间：
1989 至 1990
项目状态：
已结题

项目摘要

(1) We extracted Japanese vocabulary related to images from about 100 papers reported in the transactions of IEICE, and the word number of nouns, verbs, adjectives and adverbs we got were about 3600, 400 250 and 70, respectively.(2) We constructed an object oriented image processing system which runs on a workstation under UNIX. Besides of many fundamental lisp functions, image processing procedures and some functions realizing objective environment were built in it. We can execute both natural language processing and image processing on the system.(3) We constructed a conceptual dictionary of words directly associated with image processing. In this dictionary, concepts were categorized by conceptual upper/lower class and lower classes can inherit attributes and methods of upper ones.(4) We made up a Japanese language analysis system and applied it to a manーmachine interface of image processing. This system generates image processing functions from input Japanese sentence using the conceptual dictionary.(5) We developed a visualizing system of "room scene" which was depicted in Japanese language. Analyzing an input sentence, constraints of both objects themselves and relations among them were extracted. If ineviatable informations for the visualization, for example, shape and color of object, screen position, viewpoint and so on, were not given, default values stored in the dictionary were used.

(1)我们从IEICE学报上发表的约100篇论文中提取了与图像相关的日语词汇，得到的名词、动词、形容词和副词的词汇量分别约为3600、400、250和70个。(2)我们构造了一个在UNIX操作系统下的工作站上运行的面向对象的图像处理系统。该系统除了具有许多基本的lisp函数外，还内置了图像处理程序和一些实现客观环境的函数，既可以进行自然语言处理，也可以进行图像处理。(3)我们构建了一个与图像处理直接相关的词汇的概念词典。在该字典中，概念按概念的上/下类进行分类，并且下类可以继承上类的属性和方法。(4)我们构造了一个日语分析系统，并将其应用于图象处理的人机接口中。该系统使用概念词典从输入的日语句子生成图像处理功能。(5)我们开发了一个可视化系统的“房间场景”，这是描绘在日本的语言。通过对输入句子的分析，提取出了两个对象本身的约束条件以及它们之间的关系。如果没有给出可视化的不可避免的信息，例如对象的形状和颜色、屏幕位置、视点等，则使用存储在字典中的默认值。