基于信息融合算法的暴力视频内容识别
|
摘要
针对暴力视频的检测方法均是单一模态的且效率相对较低等问题,提出文本、视频以及音频3种模态的信息融合算法,构建网络暴力视频识别系统,设计文本以及视音频分类器;将文本分类器当做预分类器完成视频的分类操作,从而获得候选暴力影视,随之运用视音频分类器完成对其的进一步分类;开展实验分析,并与单一模态的分类算法进行对比。结果表明,该信息融合算法显著提升了识别精度,缩减了计算量,改善了分类效果。
Aiming at the problem that the detection methods of violent video were single mode and relatively inefficient, an information composite algorithm of text, video and audio modes was proposed, a network violent video recognition system was constructed, and a text and audio classifier was designed. The text classifier was used as a pre-classifier to completed video classification, so as to obtain candidate violent movies and videos, then the video and audio classifier was applied in the further classification. The experiment was carried out and compared with the single mode classification algorithms. The results show that the information fusion algorithm significantly improves the recognition accuracy, reduces the amount of calculation, and improves the classification effect.
Aiming at the problem that the detection methods of violent video were single mode and relatively inefficient, an information composite algorithm of text, video and audio modes was proposed, a network violent video recognition system was constructed, and a text and audio classifier was designed. The text classifier was used as a pre-classifier to completed video classification, so as to obtain candidate violent movies and videos, then the video and audio classifier was applied in the further classification. The experiment was carried out and compared with the single mode classification algorithms. The results show that the information fusion algorithm significantly improves the recognition accuracy, reduces the amount of calculation, and improves the classification effect.
引文
[1] 宋伟,任栋,于京,等.一种新的基于三维卷积共生梯度直方图和多示例学习的特殊视频检测算法[J].计算机学报,2018(1):1-6.
[2] LU S H,CHIANG D A,KEH H C.Chinese text classification by the Na?ve Bayes classifier and the associative classifier with multiple confidence threshold values[J].Knowledge-Based Systems,2010,23(6):598-604.
[3] JIANG J Y,LIOU R J,LEE S J.A fuzzy self-constructing feature clustering algorithm for textclassification[J].IEEE Transactions on Knowledge and Data Engineering,2011,23(3):335-349.
[4] WU S,ZHANG W,XU H.A text feature selection algorithm based on analysing therelationship between words[J].Computer Engineering & Science,2012:33-34.
[5] ZHANG Z,KAWABATA H.Study of detection method of abnormal sound using cellular neural network[J].Transactions of the Japan Society of Mechanical Engineers,2003,69(12):3207-3214.
[6] FU Z,ANTONIO R,ZHOU J.Multiple instance learning with instance selection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2011,33(5):958-977.
[7] 王静.视频分割与基于内容的自动分类:新闻故事单元检测[D].西安:西安理工大学,2005.
[8] 许荣.融合音视频特征的足球视频检索研究[D].长春:吉林大学,2008.
[9] 高新波,田春娜,张娜.一种基于 SVM 主动学习的卡通视频检测方法[J].电子与信息学报,2007,29(6):1338-1342.
[10] 宋伟,张栋梁,齐振国,等.一种基于三维卷积网络的暴力视频检测方法[J].信息网络安全,2017(12):54-60.
[11] 任栋,宋伟,于京,等.特殊视频内容检测算法研究综述[J].信息网络安全,2016(9):184-191.
[12] 刘宣文,张英萍,种聪.电视暴力内容分析及应对策略:一项基于编码技术的研究[J].杭州师范大学学报(自然科学版),2015,14(4):353-361.
[13] 李荣杰,蒋兴浩,孙锬锋.一种基于音频词袋的暴力视频分类方法[J].上海交通大学学报,2011,45(2):214-218.
[14] 高岩.基于动态与静态信息融合的暴力视频检测[D].天津:中国民航大学,2016.
[15] 王辰,张宪海,老松杨,等.基于声、像特征的视频暴力场面的探测[J].小型微型计算机系统,2001(4):456-458.
[2] LU S H,CHIANG D A,KEH H C.Chinese text classification by the Na?ve Bayes classifier and the associative classifier with multiple confidence threshold values[J].Knowledge-Based Systems,2010,23(6):598-604.
[3] JIANG J Y,LIOU R J,LEE S J.A fuzzy self-constructing feature clustering algorithm for textclassification[J].IEEE Transactions on Knowledge and Data Engineering,2011,23(3):335-349.
[4] WU S,ZHANG W,XU H.A text feature selection algorithm based on analysing therelationship between words[J].Computer Engineering & Science,2012:33-34.
[5] ZHANG Z,KAWABATA H.Study of detection method of abnormal sound using cellular neural network[J].Transactions of the Japan Society of Mechanical Engineers,2003,69(12):3207-3214.
[6] FU Z,ANTONIO R,ZHOU J.Multiple instance learning with instance selection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2011,33(5):958-977.
[7] 王静.视频分割与基于内容的自动分类:新闻故事单元检测[D].西安:西安理工大学,2005.
[8] 许荣.融合音视频特征的足球视频检索研究[D].长春:吉林大学,2008.
[9] 高新波,田春娜,张娜.一种基于 SVM 主动学习的卡通视频检测方法[J].电子与信息学报,2007,29(6):1338-1342.
[10] 宋伟,张栋梁,齐振国,等.一种基于三维卷积网络的暴力视频检测方法[J].信息网络安全,2017(12):54-60.
[11] 任栋,宋伟,于京,等.特殊视频内容检测算法研究综述[J].信息网络安全,2016(9):184-191.
[12] 刘宣文,张英萍,种聪.电视暴力内容分析及应对策略:一项基于编码技术的研究[J].杭州师范大学学报(自然科学版),2015,14(4):353-361.
[13] 李荣杰,蒋兴浩,孙锬锋.一种基于音频词袋的暴力视频分类方法[J].上海交通大学学报,2011,45(2):214-218.
[14] 高岩.基于动态与静态信息融合的暴力视频检测[D].天津:中国民航大学,2016.
[15] 王辰,张宪海,老松杨,等.基于声、像特征的视频暴力场面的探测[J].小型微型计算机系统,2001(4):456-458.