Automatic segmentation of news items in a MPEG-2 stream is a significant research topic for implementing an antomatic cataloging system of news video. This paper presents an approach which employes audio and video fea...Automatic segmentation of news items in a MPEG-2 stream is a significant research topic for implementing an antomatic cataloging system of news video. This paper presents an approach which employes audio and video feature information to automatically segment news items. Combining the analysis techniques of audio and video can overcome the weakness of the approach which only uses the image analysis techniques. This combination makes our appoach more widely adaptable to variable existence situations of news items. The proposed approach detects silence clips in accompanying audio,and integrates with shot segmentation results ,as well as anchor shot detection results ,to determine boundaries between two news items. Experimental results show that the integration of audio and video features is an effective approach to solve the problem of automatic news items segmentation.展开更多
足球视频精彩事件检测一直是视频语义分析领域研究的热点和难点.文中利用隐条件随机场(hidden conditional random field,HCRF)模型在表达和识别语义事件方面的强大功能,提出一种多维语义线索和HCRF的角球、点球和红黄牌精彩事件检测框...足球视频精彩事件检测一直是视频语义分析领域研究的热点和难点.文中利用隐条件随机场(hidden conditional random field,HCRF)模型在表达和识别语义事件方面的强大功能,提出一种多维语义线索和HCRF的角球、点球和红黄牌精彩事件检测框架.首先通过对精彩事件视频结构语义进行分析,定义了10种多维语义线索,以准确描述精彩事件富含的语义信息;然后对视频片段进行物理镜头分割,对镜头关键帧提取多维语义线索得到特征矢量,再将测试视频片段中所有镜头的特征矢量共同构成观察序列;最后在小规模训练样本的情况下将观察序列作为HCRF模型的输入,建立了精彩事件检测的HCRF模型.文中基于音视频底层特征、多维语义线索及精彩语义事件之间的映射关系,从视频结构语义的多个维度挖掘了精彩事件的内在规律,准确地实现了精彩事件的检测.实验结果表明了该框架的有效性.展开更多
文摘Automatic segmentation of news items in a MPEG-2 stream is a significant research topic for implementing an antomatic cataloging system of news video. This paper presents an approach which employes audio and video feature information to automatically segment news items. Combining the analysis techniques of audio and video can overcome the weakness of the approach which only uses the image analysis techniques. This combination makes our appoach more widely adaptable to variable existence situations of news items. The proposed approach detects silence clips in accompanying audio,and integrates with shot segmentation results ,as well as anchor shot detection results ,to determine boundaries between two news items. Experimental results show that the integration of audio and video features is an effective approach to solve the problem of automatic news items segmentation.
文摘足球视频精彩事件检测一直是视频语义分析领域研究的热点和难点.文中利用隐条件随机场(hidden conditional random field,HCRF)模型在表达和识别语义事件方面的强大功能,提出一种多维语义线索和HCRF的角球、点球和红黄牌精彩事件检测框架.首先通过对精彩事件视频结构语义进行分析,定义了10种多维语义线索,以准确描述精彩事件富含的语义信息;然后对视频片段进行物理镜头分割,对镜头关键帧提取多维语义线索得到特征矢量,再将测试视频片段中所有镜头的特征矢量共同构成观察序列;最后在小规模训练样本的情况下将观察序列作为HCRF模型的输入,建立了精彩事件检测的HCRF模型.文中基于音视频底层特征、多维语义线索及精彩语义事件之间的映射关系,从视频结构语义的多个维度挖掘了精彩事件的内在规律,准确地实现了精彩事件的检测.实验结果表明了该框架的有效性.