Image database retrival system by integrated image and text information

将图像特征与文本信息相结合，构建图像数据库检索系统。采用标准的图像处理技术，自动提取图像的轮廓线、目标形状、平均颜色信息、主要目标区域的颜色信息等特征，并存储起来作为图像检索信息。通过标准的自然语言处理技术对与图像配对的文本进行分析，提取术语词和形容词词。术语以树状结构排列，这种树状结构是由特定研究领域的结构派生出来的。这在检索中被用作一种同义词典。检索可以通过关键字组合、图像特征或这些的组合来完成。例如，“一种植物，它的果实是圆的，颜色是橙色的。”可以通过这些词来查找，也可以通过在显示的颜色表上画上适当的颜色来查找。

Image database retrieval system is constructed by integrating image features and text information. By using standard image processing techniques, many image features like contour lines, object shapes, average color information, color information of major object areas and so on are extracted automatically and are stored for image retrieval information. Texts which are paired with images are analyzed by standard natural language processing techniques, and terminological words and adjectival words are extracted. Terminological words are arranged in a tree structure which is derived by the structure of a particular study area. This is used as a kind of a thesaurus in retrieval.Retrieval can be done by keyword combination, by image feature or by the combination of these. For example, "a plant whose fruit is round and the color is orange." can be looked for either by these words or by painting proper color in a displayd color chart.