摘要随着计算机网络技术、宽带技术的不断发展和进步,近年来数字视频数据指数增长,如何快速浏览大量的视频数据以及快速的了解视频的大概内容,在很多时候尤为重要。3D-HEVC视频扩展框架已经基本成型,未来的3D视频将以基于深度的3D视频为主。因此,对3D视频关键帧摘要技术的研究会显得十分必要。43217
本文首先介绍了视频关键帧摘要技术的发展历程和特点,总结了现有的视频摘要技术的各种算法,分析了视频摘要技术的关键环节。同时,介绍了3D视频的结构,对基于深度的3D视频(包括彩色视频和对应的深度视频)提取关键帧以帮助正常彩色视频的关键帧摘要。关键帧摘要主要是镜头分割和关键帧提取两个步骤,要根据图像的物理特征以及背景来表征此帧图像的内容。本文比较相邻两帧图像的灰度直方图的相似性,然后通过自适应阈值算法得出镜头分割阈值T1和局部关键帧提取阈值T2,之后根据阈值T1进行镜头分割,然后根据镜头内阈值T2提取关键帧形成视频摘要。
此外,文章对3D视频的彩色视频和深度视频分别提取关键帧摘要,结果表明深度视频的关键帧摘要有助于彩色视频的关键摘要。最后,本文对未来视频摘要技术的发展以及应用做出了展望。
关键词 3D视频摘要 镜头分割 关键帧提取 灰度直方图
毕业论文设计说明书外文摘要
Title Research on Key Frames Extraction Technology of 3D Video
Abstract
With the rapid development and progress of computer network technology, as well as broadband technology, digital video data comes out an exponential growth. It is particularly important to browse a large amount of video data quickly and understand the video content quickly. Besides, the 3D-HEVC video framework has been basically formed with video-plus-depth format. Therefore, it is necessary to study the video summarization technology in the coming 3D video format.
In this paper, we first introduce the history and characteristics of key frame extraction technology. Existing video abstract technologies algorithms are summarized, and the key idea of these video abstract algorithms are analyzed. Then, this paper introduces the structure of 3D video, and extracts key frames based on the depth of the 3D video, including color video and the corresponding depth video, to help the normal color video extracts key frames. The key frame abstraction is pided into two stages: shot segmentation and key frame extraction. The physical characteristics of the image and background are utilized to characterize the contents of this frame image. We compares the similarity between two gray histogram of adjacent images, then use adaptive threshold algorithm to get the threshold for shot segmentation and local key frame extraction threshold. The two thresholds are utilized to carry out shot segmentation and key frames extraction to compose a video summarization.
In addition, We try to extract the key frames of 3D video (the color video plus depth video). The results show that the depth 3D video’s key frame abstraction is helpful to color video’s summarization. Finally, the future work of video abstraction technology is introduced .
Key words 3D video abstract shot segmentation key frame extraction gray histogram
目 次
第1章 绪论 1
1.1 研究背景 1
1.2 视频摘要的研究现状 1
1.3 视频摘要一般的研究内容 2
1.4 本文的研究内容以及论文组织 3
第2章 视频摘要的相关概念及技术