硕士学位论文-基于纹理和形状检索的视频信息管理系统的研究与实现汇.doc

摘 要 基于内容的视频检索技术是当前研究的热点问题。它主要是利用视觉特征，如颜色、纹理、形状和空间关系等特征来进行检索。这种检索方式突破了传统的基于文本检索技术的局限，直接对视频内容进行分析并抽取特征，然后利用这些内容特征建立索引并进行检索。 本文首先详细地阐述了基于内容的视频检索技术的原理、关键技术和研究现状，并对基于纹理和形状检索的视频信息管理系统进行了分析。 本文的重点是研究了视频图像中的纹理和形状特征的提取算法和检索方法，对灰度共生矩阵特征描述参量的物理意义进行了阐释的查率和查准率进行了统计At present, Content-Based Video Retrieval (CBVR) is becoming a hot research topic. It is a retrieval technology based on the vision features, such as the color, texture and shape. This retrieval method breakthrough the limitation of traditional one based on text. It analyzes and extracts features of video content directly, then uses these features to establish a retrieve system as an index. Firstly, we describe the principle, essential technology, and the present research situation of the content-based video retrieval system in detail, and then analyze the texture and shape based video retrieval system. We lay a strong emphasis on the research of texture and shape character pick-up algorithm and retrieval technology, explain the meaning of the parameters of Gray-Level Co-occurrence Matrix, and compare the result of three algorithm —Canny, Laplacian and Sobel. We select Canny and Gray-Level Co-occurrence Matrix in this system. We implement a shape and texture based video information management system with SQL Server and Visual C++, evaluate the performance of it. The experiments show that the performance of the retrieval system based on texture and shape is relative good especially in searching the image with typical texture and shape character. Key words：Shape; Texture; Content-Based Video Retrieval; Moment Invariant; Gray-Level Co-occurrence Matrix 第一章 绪 论 1 1.1课题的研究背景 1 1.2课题的研究内容及成果 3 第二章 基于内容的视频检索技术 4 2．1 基于内容的视频检索概述 4 2．2 基于内容的视频检索的关键技术和研究现状 5 2.2.1基于内容的视频检索的关键技术 5 2.2.2 基于内容的视频检索技术研究现状 7 第三章 纹理和形状特征的算法选择 9 3．1纹理特征的选择 9 3.1.1 纹理描述 9 3.1.2 纹理特征 12 Tamura 纹理特征 12 自回归纹理模型 14 灰度共生矩阵 15 3.1.3 相似性度量技术 17 3．2形状特征的选择 18 3.2.1 图像边缘检测技术 18 边缘检测原理和分类 18 梯度算子边缘检测算法 19 Canny边缘检测算法 21 几种边缘检测算法的比较 22 3.2.2 形状特征描述 23 矩的概