基于纹理特征的图像分类研究-计算机应用技术专业论文.docxVIP

- - I - 摘 要 随着科技的发展，多媒体数据库里的图像信息越来越多，并且增长速度越来越 快，它在提供大量资料的同时给我们带来了烦恼：如何对如此大的数据库进行整理 与分类便于用户进行检索。 多媒体数据库的数据量如此巨大，给我们检索数据带来了巨大的挑战，依靠传 统的人工注释的方法已经越来越无法满足用户的需求。近年来，基于内容的图像检 索技术(CBIR)有了长足的发展。基于内容的图像检索以人的视觉感受对图像进行检 索，它对图像的内容进行分析，并从中提取特征进行描述，然后利用图像的内容特 征建立检索，并最终完成检索图像，使得检索过程更加准确与高效。 检索图像的内容包括图像的颜色、纹理、形状等视觉特征和语义特征等，其中， 纹理是最为显著的视觉特征之一。由于纹理的描述和分析的复杂性，在基于内容的 图像检索系统中纹理特征还没有得到充分利用。本文主要对纹理分割和基于纹理特 征的图像分类进行了研究。 纹理分割是纹理图像分析的关键步骤，因为只有完成对图像的分割才能提取图 像中对象的视觉特征和语义特征。论文讨论了纹理分割的一般方法，对基于统计的 灰度共生矩阵方法进行了研究，分析了距离、灰度级和窗口大小对特征提取和分割 效果的影响。采用灰度共生矩阵方法提取纹理特征，利用 K 均值聚类方法进行了纹 理分割，取得了比较满意的实验结果。 关键词：图像分类；图像检索；纹理分割；K-means - - II - Abstract With technological development, the information content in multimedia database is more and more larger and the speed turns more and more quickly. It provides a lot of information also to us. However, it also brings a lot of troubles to us. The problem is how to sort and classify information in such a database. Multimedia database is so large that we retrieve data become more difficult. Rely on traditional methods is not meet users needs. In recent years, Content-based Image Retrieval (CBIR) has made great significant progress. It analyses the content of the image and gets characteristic. Then it uses image feature to build the indexing to retrieval image. That retrieves more effective. CBIR retrieves the content’s color, texture, shape, and other visual features. Texture is the most notable feature of the visual feature. Texture features does not get the full use because of its difficulty. This article mainly research texture splitting and CBIR. Splitting texture is the key steps to analyses the image. We get the feature of the visual feature after slitting image. This article discusses the general approach to analyses the division of texture. This article discusses GLGM (the grey level grows matrix) and research on the effect of distance and grey level. It uses GLGM to get texture features and use K-means algorithm to slit texture. It achieved a satisfactory r