[1]李强军,李启南.基于互信息量和自回归模型的镜头分割方法[J].计算机技术与发展,2019,29(01):35-39.[doi:10. 3969 / j. issn. 1673-629X. 2019. 01. 008]
　LI Qiang-jun,LI Qi-nan.A Lens Segmentation Method Based on Mutual Information andSelf-regression Model[J].,2019,29(01):35-39.[doi:10. 3969 / j. issn. 1673-629X. 2019. 01. 008]
点击复制

基于互信息量和自回归模型的镜头分割方法()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
29
期数:
2019年01期
页码:
35-39
栏目:
智能、算法、系统工程
出版日期:
2019-01-10

文章信息/Info

Title:
A Lens Segmentation Method Based on Mutual Information andSelf-regression Model
文章编号:
1673-629X(2019)01-0035-05
作者:
李强军;  李启南
兰州交通大学 电子与信息工程学院,甘肃 兰州,730070
Author(s):
LI Qiang-jun; LI Qi-nan
School of Electrical and Information Engineering,Lanzhou Jiaotong University,Lanzhou 730070,China
关键词:
镜头分割;  自适应阈值;  互信息量;  自回归模型
Keywords:
lens segmentation; adaptive threshold; mutual information; autoregressive model
分类号:
TP301. 6
DOI:
10. 3969 / j. issn. 1673-629X. 2019. 01. 008
摘要:
随着互联网的急速发展,盗版,不健康,暴力等视频在网络上肆意流窜,如何快速、有效、准确地对视频数据进行管理,已然成为迫切需要解决的问题.在视频处理过程中,首先是对镜头进行分割,然后再进行视频帧的分析处理.然而许多视频内容的复杂性比较高,不一定能得到比较好的分割结果.鉴于此,提出一种基于互信息量和自回归模型的自适应阈值镜头分割算法.该算法首先以非均匀分块加权HSV直方图为基础,通过计算两帧的互信息量求出两帧的相似度值,然后建立自回归模型产生自适应阈值进行镜头分割,最终实现突变镜头的突变检测和渐变镜头的渐变检测,并采用时间窗口进一步降低检测误差.以优酷网上随机抽取下载的真实视频为测试对象,实验结果表明,该算法可适用于不同类型的视频镜头分割,具有很好的检测效果.
Abstract:
With the rapid development of Internet,piracy,unhealthy,violence and other videos flows freely on the Internet. How to manage video data quickly,effectively and accurately has become an urgent problem to be solved. In the video processing,the lens is firstsegmented,and then the analysis and processing of video frames are carried out. However,the complexity of many video contents is relatively high,so it is not necessary to get a better segmentation result. In view of this,we propose an adaptive threshold lens segmentationalgorithm based on mutual information and self-regression model. Firstly based on the non-uniform block weighted HSV histogram,through calculating the mutual information of two frames the similarity of two frames is computed. And then a regression model is established to generate adaptive threshold for lens segmentation. Finally,the mutation detection of mutation lens and gradient detection of gradient lens are realized,and the time window is used to further reduce the detection error. The real video downloaded randomly on Youkuis selected as the test object. The experiment shows that the proposed algorithm is applicable to different video lens segmentation withbetter detection effect.

相似文献/References:

[1]汪翔 罗斌 翟素兰[] 涂铮铮.基于颜色空间的自适应阈值镜头分割算法[J].计算机技术与发展,2012,(09):37.
　WANG Xiang,LUO Bin,ZHAI Su-lan,et al.Self-threshold Shot Segmentation Based on Color Space[J].,2012,(01):37.
[2]夏磊,张明. 篮球赛事视频镜头分割[J].计算机技术与发展,2015,25(02):216.
　XIA Lei,ZHANG Ming. Shot Segmentation of Basketball Tournament Video[J].,2015,25(01):216.
[3]靳冰凌,张震,张子耀.基于视觉的驾驶员疲劳特征提取方法[J].计算机技术与发展,2018,28(11):193.[doi:10.3969/j.issn.1673-629X.2018.11.042]
　JIN Bing-ling,ZHANG Zhen,ZHANG Zi-yao.A Driving Fatigue Feature Extraction Method Based on Vision[J].,2018,28(01):193.[doi:10.3969/j.issn.1673-629X.2018.11.042]

更新日期/Last Update: 2019-01-10