Title :
Object based video synopsis
Author :
Tong Yao ; Maosen Xiao ; Caiwen Ma ; Chao Shen ; Peng Li
Author_Institution :
Opt. Direction & Pointing Tech. Res. Dept., Xi´an Inst. of Opt. & Precision Mech., Xi´an, China
Abstract :
Tens of thousands of surveillance cameras all over every corner of the world send large amounts of data 24 hours a day, but most of these videos are redundant data lacking of activity. It is time consuming to browse all of those raw data. This paper has researched object based video synopsis technology with the surveillance video of a neighborhood and an indoor surveillance video as research target, presenting all necessary activities in a short time by displaying objects concurrently, even if they originally occurred at different time. The result shows that we can create a synopsis with maximum activity while avoiding conflicts between objects as much as possible.
Keywords :
video signal processing; video surveillance; indoor surveillance video; neighborhood video; object based video synopsis technology; surveillance cameras; Computer vision; Object detection; Object tracking; Pattern recognition; Sparse matrices; Surveillance; Target tracking; background modeling; multi-target tracking; object detection; video synopsis;
Conference_Titel :
Advanced Research and Technology in Industry Applications (WARTIA), 2014 IEEE Workshop on
Conference_Location :
Ottawa, ON
DOI :
10.1109/WARTIA.2014.6976479
