![](bg4.png)
background subtraction such as Codebook Construc-
tion (Kim et al., 2004) can be used. Finally, tech-
niques based on optical flow estimation (Horn and
Schunck, 1993) overcome the shortcomings of a fixed
background.
Furthermore, in order to protect the sensitive in-
formation and, at the same time, allow its disclosure
under certain conditions (e.g. for criminal investi-
gation), invertible operations/functions must be used.
The so-called coefficient alteration techniques (Du-
Faux and Ebrahimi, 2006) are invertible operations in
which the process of protection (and disclosure) de-
pends on a key. The key is used to generate a “pro-
tection stream” – a set of bits that will be used to
specify how the information in the regions of inter-
est is changed. Since this technique alters the sign of
the coefficients of the compressed video streams, its
use does not affect the efficiency of the compressor
(which is essential in any video storage network). Au-
thorities could cryptographically generate the “pro-
tection stream” whenever it is necessary to use it to
disclose a portion of the protected video.
5 CAMNOTIFY: MAKING VSS
VISIBLE
The right of people to privacy in VSS can be com-
plemented with the ability to easily know the cam-
eras of VSS that have recorded a given person. Hence
we propose a website database of VSS and a comple-
menting mobile application (CamNotify) that might
help people realise about the privacy concerns related
to pervasive video surveillance. The VSS using our
software packages (allowing accurate and real-time
video protection) should be listed in the database so
citizens, equipped with a mobile phone running our
app, could be notified of the nearby VSS.
Note that by means of this application, users could
collaborate with the web service in different ways,
for instance, by adding comments about the VSS, by
voting, and by providing information about possible
abuse or misuse. In that sense, collaborative filtering
techniques could be also applied over the data stored
in the census database of the service to obtain a holis-
tic view of the privacy of users with regard to VSS.
Finally, users could notify the existence of cameras
belonging to VSS that are not using the privacy pro-
tection techniques. Consequently, the VSS database
could be updated collaboratively by the users of the
application.
We have designed and implemented a prototype
of this mobile application that is described in the fol-
lowing lines. With the aim to illustrate the scenario,
let us consider the following example: Peter Smith
enrols in the web service to be aware of the VSS that
might have recorded him during his everyday activi-
ties. After downloading the CamNotify application to
his smartphone, he registers to the service and turns
on the application. The mobile app uses the self-
location and communication capabilities of the device
to request the web service a list of nearby cameras, ac-
cording to its current position (longitude and latitude).
In fact, this action occurs each time the smartphone
detects a significant change in Peter’s location.
Now, the app is able to measure distances between
the current location of the device and the location of
the cameras in the list. The cameras located close to
the smartphone (i.e. that might have been recording
Peter) will be listed in the display (on a map). Also,
the list of cameras is shown 4. In addition, this list of
cameras will be sent to the web service, which will be
also aware of the time in which Peter passed nearby
each camera.
After some time, Peter enters a shop willing to buy
some coffee. He realises that the shop owners have
installed several cameras in the shop premises. How-
ever, there is no visible sign advertising them nor any
display in which the recorded video is shown. Peter
takes a look at his smartphone and realises that this
VSS is not in the list. Hence, he makes use of Cam-
Notify to request the addition of a new camera in that
location 3. Certainly, he will specify a low trust level
for the VSS.
When he arrives home, Peter opens his laptop and
connects to the web service website (cf. Figure 2)
and he can access the list of cameras that might have
recorded him, which are grouped by walks. He can
read the comments of other users, manage some in-
formation and, naturally, he can make comments on
the cameras he saw during the day (for instance, by
warning that a camera that is currently labelled with
a high level of trust, actually does not have a visible
display).
Two hours later, Mary Martin, who is also a Cam-
Nofity user, enters the shop previously visited by Pe-
ter. She receives an alarm in the application: Cam-
Notify warns her that there is a new camera being de-
tected nearby and the web service needs to confirm
that there is a real camera there (note that the cam-
era was included in the system a few hours ago and it
could be a fake). Mary confirms with a positive vote
that the camera is actually there. Naturally, after a
certain number of positive votes, the web service will
stop requesting CamNotify users for confirmations.
However, CamNofity can randomly request users to
approve cameras from time to time, aiming at being
notified of VSS not operating any more.
PECCS2014-InternationalConferenceonPervasiveandEmbeddedComputingandCommunicationSystems
52