Game Development Reference
In-Depth Information
metro entries. In the recent Boston Marathon bombing case, the two suspects were
identified from surveillance videos via a dramatically human-led endeavor.
Reduced network load and storage
. IVS that includes video event detection can
minimize the need for storage space by recording only video that contains activity.
Furthermore, by processing as much as possible of the video in the network cameras
themselves, the load on the network is significantly reduced as only relevant video is
streamed from the cameras, which improves the efficiency of the other video analysis
applications.
More intelligent recommendation service
. By analyzing the streamed video,
IVS can provide more useful recommendations or suggestions. For example, IVS
can provide shortest path recommendation by using traffic jam analysis.
In a word, the intelligent video system is a promising solution for the data tsunami
of surveillance videos. It makes the cameras in surveillance system not only as
“eye” to recode image about scene, but also as “brain” to think about the scene by
finding the meaningful events for us. There are two broad categories of systems for
implementing intelligent video systemcentralized and distributed (Axis
2009
). In
centralized architectures, video and other information are collected by cameras and
sensors and brought to a centralized server for analysis, which is denoted as intelligent
video analysis system in this topic. In distributed architectures, the edge devices
(network cameras and video encoders) are “intelligent” and can extract relevant
basic perceptual or semanteme information. The following sections will introduce
the two implementations for intelligent video system, respectively.
10.2 Intelligent Video Analysis System
Traditional video analysis system is centralized system, which extracts interesting
objects and events from the compressed nonstructural video data to reorganize them
as structural data. And then many computer vision algorithms, e.g., pedestrian detec-
tion, face recognition, event detection, are integrated to jointly finish some intelligent
tasks. This kind of intelligent video system mainly focuses on the video analysis.
10.2.1 Framework of Intelligent Video Analysis System
Since the ultimate goal of intelligent video for security and surveillance is to auto-
matically detect events and situations that require the attention of security personnel,
it needs many smart analysis modules to jointly support the intelligent system, which
is the core part IVS. In addition, for the video analysis system, it also needs the video
collection part by cameras or sensors, and the human-computer interface part, which
provides intelligent operations or suggestions to people, and receives commands from
operators. In this architecture, the collected video data from different cameras are
