matlab基于视觉的多媒体描述系统设计与实现
时间:2018-07-18 20:51 来源:毕业论文 作者:毕业论文 点击:次
| 摘要随着互联网的快速发展，以视频数据为代表的多媒体数据呈现出了海量增长趋势。为了对数据量庞大的数字视频进行管理，视频语义分析技术由此提出。 本文首先对视频进行了结构化分析，涉及到了镜头边缘检测、关键帧提取和镜头分类三部分内容。本文采用了一种自适应的双阈值镜头边缘检测算法。通过图像熵值来确定关键帧。同时，本文使用了一种结合比赛特征的镜头分类算法，成功地将所有镜头细分为球场全局镜头、球员特写镜头和观众镜头三类。 通过提取图像的最大极值稳定区域，对字幕区域进行了准确定位，并识别出了字幕文本。对于一个比赛镜头，通过背景建模得出球场背景图片，背景差分提取运动目标并进行跟踪。采用matlab R2014作为开发工具，完成了一个网球视频的多媒体描述系统。25868 关键词 视频语义分析 镜头边缘检测 字符识别 背景差分 毕业设计说明书（毕业论文）外文摘要 Title Design and implementation of Multimedia description system based on vision Abstract With the rapid development of Internet, multimedia data, especially video data, showing the massive growth trend. To make management, analysis and data retrieval of the huge data of digital video , video semantic analysis technology based on the content is introduced in this paper. Video structure analysis is the key and foundation of the whole video semantic.This part includes the shot boundary detection, key frame extraction, shot classification and other subtasks.This paper adopts the adaptive double threshold edge detection algorithm.The key frame is chosen by using image entropy to determine the importance of all image frames in a shot.This paper constructs a shot classification algorithm ,which classifies successfully the shots into three categories:field lens, players close-up and audiences lens. By extracting the maximally stable extremal regions of image frame, this paper accurately position the caption area, and complete OCR character recognition. For a tennis playing shot, background picture can be obtained by background modeling , this paper uses background differencing to extract moving targets and tracking. Finally, this paper takes matlab R2014 as the development platform, and implement a tennis video analysis and description system. Keywords video semantic analysis shot boundary detection character recognition background difference 目 录 第1章 绪论 1 1.1 引言 1 1.2 国内外的发展及现状 1 1.3 本文的工作 2 1.4 本文组织安排 3 第2章 视频结构化分析与研究 3 2.1 镜头边缘检测 3 2.2 镜头边缘检测算法简介及实现 4 2.3 关键帧提取 8 2.4 镜头分类 10 2.4.1 主色及主色率的提取 10 2.4.2 Hough直线变换检测场地线 10 2.4.3 基于主色率、网球球场边界线及图像边缘像素特征的镜头分类算法 10 2.5 本章小结 13 第3章 视频字幕检测与识别 14 3.1 视频字幕文字 14 (责任编辑：qin) |