Full-Text Search:
Home|Journal Papers|About CNKI|User Service|FAQ|Contact Us|中文
《Journal of Test and Measurement Technology》 2019-03
Add to Favorite Get Latest Update

Convolutional Two-Stream Neural Network Fusion for Video Concentration Based on Interactive Mechanism

ZHAO Chunfei;ZHANG Lihong;College of Physics and Electronic Engineering,Shanxi University;  
It is difficult to accurately extract moving targets when the mutual occlusion and background complexity between moving targets during video concentration which cause the concentration ratio to decrease.Aiming at this problem,a convolutional dual-stream fusion network is proposed which for video concentration based on interaction mechanism.Firstly,the region of interest is selected in the input video frame;then,the moving target features and the background features are extracted by the convolutional two-stream fusion network,and the features are merged to reduce the influence of mutual occlusion between the moving targets;finally,correlation operation of the fusion features are carried out through the interaction mechanism,which can effectively improve the correlation between the moving targets between the moving targets and the background,and the scene frames are clustered based on similar matrix to obtain key frames.The experimental results show that when the video is condensed through the network structure,the concentration ratio and the recall rate both improved.
【Fund】: 山西省科技攻关计划(工业)资助项目(2015031003-1)
【CateGory Index】: TP391.41;TP183
Download(CAJ format) Download(PDF format)
CAJViewer7.0 supports all the CNKI file formats; AdobeReader only supports the PDF format.
©2006 Tsinghua Tongfang Knowledge Network Technology Co., Ltd.(Beijing)(TTKN) All rights reserved