|
|
|
|
Project Name :
|
Reliable Digital Video Surveillance Systems with Efficient Content-based Video Retrieval Support (UIM/167)
|
|
|
Objective :
|
MultiVision Intelligent Surveillance Limited (abbreviated as MultiVision hereafter) is a
company specialized in the production and distribution of digital video surveillance products by
providing functions such as digital recording, real-time remote monitoring and management.
Their products offer more capabilities than the conventional analogy CCTV systems. Over the
years, MultiVision has invested heavily in R&D to improve their digital video surveillance
products.
MultiVision is now confronting two technological challenges in digital video surveillance: (a) to
perform timely and accurate video retrieval; and (b) to improve the reliability of hard disks
under high-frequency read-write environment. The objective of this proposal is to tackle these
two areas, which is crucial in improving MultiVision’s products to foster MultiVision’s
competitiveness in the digital video surveillance market.
(a) Timely and accurate video retrieval
Due to the decrease in storage cost and the advances in video compression technologies, a
typical digital video surveillance system can now store at least 7 days of video per camera on
a single hard disk, resulting in terabytes of video data. With this large amount of video data,
the browsing and searching for specific video of interest is an extremely tedious task for
human. It is therefore necessary to equip video surveillance products with efficient and
accurate video retrieval functions such that users can search for specific videos of interest,
within reasonable time, and with minimal amount of human intervention. This is, however,
an extraordinary challenging task because there is no commonly agreed method for analyzing and matching video contents such that video segments can be matched against a
user specified retrieval query. Furthermore, even the appropriate tools for analyzing video
content exist, it remains hard to identify the relevant information for indexing such that the
interested video can be retrieved efficiently. It is therefore necessary to research into the
problem of video retrieval such that browsing and search for specific videos of interest can
be performed in an accurate and timely manner.
The first objective of this proposal is to develop a video retrieval and indexing scheme such
that it can be deployed in MultiVision’s product to fulfill the anticipated demand of video
retrieval function from clients.
(b) Improving the reliability of hard disks under high-frequency read-write environment
A digital video recording server produced by MultiVision can record 16 video feeds
simultaneously, meaning that there is a total of 16 video data writers competing for hard disk
write operations. Besides, the recording server also needs to read video data from hard disk whenever the user requests for a specific video playback or during the video data backup
process. The result is that the hard disk in a video server is being operated under
high-frequency read-write environment. Furthermore, as outdated video will be flushed out
from the disk, the multiple writers environment could eventually result in severe
fragmentation problem in the hard disk. As a consequence, the hard disk in a video recording
server is subjected to work under both high data rate and pseudo random high-frequency
read-write environment. This shortened the life of a hard disk and hence frequent
replacement of hard disk is necessary, which induces huge amount of maintenance cost.
The second objective of this proposal is to develop a method for storing real-time video data
from multiple cameras which can improve the reliability of hard disks under this
high-frequency read-write environment.
|
|
|
Project Duration :
|
15 months
|
|
|
|
|
Introduction