基于内容的图像检索技术研究-控制科学与工程专业论文.docxVIP

摘要摘要 摘要 摘要 随着计算机科学技术的发展、Internet发展和应用的普及以及多媒体数据库的 迅速增加，我们拥有了海量的数字图像和视频信息。只有合理的组织这些资源， 并且高效的查询和检索方法，才有可能充分使用这些资源。本文在基于内容的彩 色图像检索算法等方面展开研究。主要分为以下几个方面的内容： 1．本文提出了一种基于CL多小波分析的图像检索方法。多小波具有一些单 小波所没有的优点，如多小波可以集紧支性、正交、对称和消失矩于一身， 应用于图像处理方面可显现出很大的优势。该方法可以更精确的利用多个 小波基之间的关系进而反映图像高低维信息之间的关系，提高图像检索系 统的性能。 2．提出了基于分块主色和纹理特征的图像检索算法和一种综合主色调及形 状特征的图像检索算法，算法采用分块主色提取作为主颜色特征的提取方 法，有着计算量小，易于实现的优点。 3．介绍了一种图像特征提取的进化自组织代理的方法；从进化自组织代理的 观点，正式阐明了从一幅图像中进行特征提取的全局最优标准：叙述了自 组织代理的表示法，他们的互反应行为，以及基于代理的图像特征提取的 运算法则。 关键词： CL多小波变换图像检索特征提取相似度代理 ABSTRACTABSTRACT ABSTRACT ABSTRACT With the more and more extensive use of multimedia database，especially a great lot of image database，effective image retrieval has attracted quite a number of people’S attention．Beyond question，CBIR has many advantages in resolving the problems we are facing．Content Based Color Image Retrieval(CBIR)is studied in this dissertation．The main contributions Can be summarized as follows： 1．CL multi-wavelet decomposition based image retrieval algorithm is proposed， and much more accurate relationships of some wavelet coefficients can be made use of tO reflect the relationships between the details and the general picture of the image．To begin with，multi-wavelet decomposition method is used because of its merits，such as simultaneously orthogonal，symmetric，short supported and of high vanishing moments，and it has many advantages in image processing also． The performances ofthe CBIR system can be improved with the method． 2．In the research of CBIR，the bottom features，such as colors，texture and area shape，etc，are the main objects to be analyzed．Because the retrieval effect is not well with any one of them，two methods based on the integrated features are proposed．One is dominant colors of partitions and texture features based image retrieval，another is based on dominant colors and shape features．Color features Can be extracted from the ones of partitions in the algorithm．with a little calculation and easy realization． 3．An Evolutionary Autonomo